diff --git "a/vocab.json" "b/vocab.json"
new file mode 100644--- /dev/null
+++ "b/vocab.json"
@@ -0,0 +1,69791 @@
+{
+  "abi": {
+    "'": 49,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 12,
+    "c": 34,
+    "d": 30,
+    "e": 13,
+    "f": 19,
+    "g": 35,
+    "h": 36,
+    "i": 18,
+    "j": 29,
+    "k": 11,
+    "l": 14,
+    "m": 15,
+    "n": 5,
+    "o": 22,
+    "p": 21,
+    "r": 20,
+    "s": 26,
+    "t": 16,
+    "u": 17,
+    "w": 28,
+    "y": 23,
+    "|": 4,
+    "á": 40,
+    "â": 42,
+    "é": 39,
+    "ê": 41,
+    "í": 38,
+    "î": 43,
+    "ó": 44,
+    "ô": 55,
+    "ú": 46,
+    "û": 47,
+    "ě": 45,
+    "ń": 53,
+    "ǎ": 52,
+    "ǐ": 48,
+    "ǒ": 51,
+    "ǔ": 50,
+    "ɔ": 10,
+    "ɛ": 6,
+    "ɩ": 7,
+    "ʋ": 8,
+    "́": 33,
+    "̂": 32,
+    "̌": 37,
+    "έ": 27,
+    "ί": 25,
+    "ε": 24,
+    "ḿ": 54
+  },
+  "abk": {
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "а": 4,
+    "б": 27,
+    "в": 45,
+    "г": 18,
+    "д": 23,
+    "е": 19,
+    "ж": 36,
+    "з": 14,
+    "и": 7,
+    "к": 25,
+    "л": 17,
+    "м": 20,
+    "н": 11,
+    "о": 22,
+    "п": 33,
+    "р": 9,
+    "с": 15,
+    "т": 21,
+    "у": 10,
+    "ф": 44,
+    "х": 13,
+    "ц": 29,
+    "ч": 40,
+    "ш": 16,
+    "ы": 6,
+    "ь": 12,
+    "џ": 42,
+    "қ": 32,
+    "ҟ": 30,
+    "ҩ": 35,
+    "ҭ": 26,
+    "ҳ": 24,
+    "ҵ": 31,
+    "ҷ": 43,
+    "ҽ": 39,
+    "ҿ": 37,
+    "ә": 8,
+    "ӡ": 34,
+    "ӷ": 38,
+    "ԥ": 28
+  },
+  "abp": {
+    "'": 34,
+    "-": 23,
+    "6": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 27,
+    "d": 19,
+    "e": 13,
+    "f": 28,
+    "g": 16,
+    "h": 10,
+    "i": 7,
+    "j": 24,
+    "k": 11,
+    "l": 14,
+    "m": 15,
+    "n": 6,
+    "o": 8,
+    "p": 17,
+    "q": 31,
+    "r": 25,
+    "s": 22,
+    "t": 12,
+    "u": 20,
+    "v": 30,
+    "w": 21,
+    "x": 33,
+    "y": 9,
+    "z": 29,
+    "|": 5,
+    "ā": 26,
+    "ō": 32
+  },
+  "aca": {
+    "'": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 9,
+    "d": 26,
+    "e": 10,
+    "f": 33,
+    "g": 32,
+    "h": 22,
+    "i": 6,
+    "j": 15,
+    "l": 12,
+    "m": 18,
+    "n": 7,
+    "o": 20,
+    "p": 30,
+    "q": 28,
+    "r": 13,
+    "s": 16,
+    "t": 23,
+    "u": 11,
+    "v": 35,
+    "w": 14,
+    "x": 36,
+    "y": 24,
+    "z": 34,
+    "|": 5,
+    "á": 8,
+    "é": 19,
+    "í": 17,
+    "ñ": 37,
+    "ó": 29,
+    "ú": 27,
+    "—": 31
+  },
+  "acd": {
+    "'": 27,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 30,
+    "d": 21,
+    "e": 13,
+    "f": 23,
+    "g": 15,
+    "h": 29,
+    "i": 9,
+    "k": 17,
+    "l": 20,
+    "m": 11,
+    "n": 7,
+    "o": 10,
+    "p": 26,
+    "r": 19,
+    "s": 16,
+    "t": 25,
+    "u": 14,
+    "w": 22,
+    "y": 12,
+    "|": 4,
+    "ŋ": 28,
+    "ɔ": 8,
+    "ɛ": 6
+  },
+  "ace": {
+    "'": 35,
+    "-": 28,
+    "0": 39,
+    "1": 43,
+    "2": 44,
+    "3": 42,
+    "4": 41,
+    "6": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 38,
+    "a": 5,
+    "b": 15,
+    "c": 29,
+    "d": 23,
+    "e": 7,
+    "f": 33,
+    "g": 9,
+    "h": 11,
+    "i": 12,
+    "j": 22,
+    "k": 13,
+    "l": 18,
+    "m": 14,
+    "n": 6,
+    "o": 16,
+    "p": 21,
+    "q": 36,
+    "r": 19,
+    "s": 20,
+    "t": 10,
+    "u": 8,
+    "w": 25,
+    "y": 17,
+    "z": 34,
+    "|": 4,
+    "á": 31,
+    "é": 24,
+    "ë": 27,
+    "ó": 32,
+    "ô": 26,
+    "ö": 30,
+    "ú": 37
+  },
+  "acf": {
+    "'": 35,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 32,
+    "d": 20,
+    "e": 22,
+    "f": 28,
+    "g": 31,
+    "h": 30,
+    "i": 8,
+    "j": 25,
+    "k": 12,
+    "l": 15,
+    "m": 17,
+    "n": 6,
+    "o": 9,
+    "p": 14,
+    "r": 33,
+    "s": 10,
+    "t": 11,
+    "u": 19,
+    "v": 24,
+    "w": 16,
+    "y": 13,
+    "z": 27,
+    "|": 4,
+    "è": 18,
+    "é": 7,
+    "ò": 26,
+    "ʼ": 29,
+    "—": 34
+  },
+  "ach": {
+    "'": 27,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 19,
+    "c": 17,
+    "d": 18,
+    "e": 8,
+    "f": 29,
+    "g": 20,
+    "h": 30,
+    "i": 7,
+    "j": 24,
+    "k": 9,
+    "l": 16,
+    "m": 11,
+    "n": 10,
+    "o": 5,
+    "p": 22,
+    "r": 21,
+    "s": 26,
+    "t": 13,
+    "u": 14,
+    "v": 28,
+    "w": 12,
+    "y": 15,
+    "|": 4,
+    "ŋ": 23
+  },
+  "acn": {
+    "-": 30,
+    "0": 32,
+    "1": 34,
+    "2": 37,
+    "3": 38,
+    "4": 33,
+    "5": 39,
+    "6": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 24,
+    "c": 26,
+    "d": 18,
+    "e": 13,
+    "f": 31,
+    "g": 11,
+    "h": 5,
+    "i": 12,
+    "j": 22,
+    "k": 20,
+    "l": 15,
+    "m": 16,
+    "n": 8,
+    "o": 9,
+    "p": 25,
+    "q": 19,
+    "r": 29,
+    "s": 7,
+    "t": 17,
+    "u": 14,
+    "v": 21,
+    "w": 27,
+    "x": 28,
+    "y": 10,
+    "z": 23,
+    "|": 4,
+    "—": 36
+  },
+  "acr": {
+    "'": 7,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 15,
+    "d": 28,
+    "e": 11,
+    "f": 35,
+    "g": 31,
+    "h": 17,
+    "i": 6,
+    "j": 12,
+    "k": 8,
+    "l": 16,
+    "m": 22,
+    "n": 13,
+    "o": 14,
+    "p": 26,
+    "q": 19,
+    "r": 9,
+    "s": 24,
+    "t": 18,
+    "u": 10,
+    "v": 37,
+    "w": 20,
+    "x": 21,
+    "y": 25,
+    "z": 27,
+    "|": 4,
+    "á": 32,
+    "é": 33,
+    "í": 36,
+    "ñ": 39,
+    "ó": 34,
+    "ú": 30,
+    "—": 29
+  },
+  "acu": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 30,
+    "c": 17,
+    "d": 26,
+    "e": 14,
+    "f": 31,
+    "g": 21,
+    "h": 15,
+    "i": 6,
+    "j": 16,
+    "k": 12,
+    "l": 27,
+    "m": 11,
+    "n": 7,
+    "o": 22,
+    "p": 20,
+    "q": 35,
+    "r": 10,
+    "s": 13,
+    "t": 9,
+    "u": 8,
+    "v": 32,
+    "w": 18,
+    "x": 36,
+    "y": 19,
+    "z": 34,
+    "|": 5,
+    "á": 29,
+    "é": 28,
+    "í": 23,
+    "ñ": 37,
+    "ó": 33,
+    "ú": 25,
+    "—": 24
+  },
+  "ade": {
+    "'": 41,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 42,
+    "d": 22,
+    "e": 8,
+    "f": 24,
+    "g": 10,
+    "h": 40,
+    "i": 12,
+    "k": 17,
+    "l": 20,
+    "m": 23,
+    "n": 7,
+    "o": 18,
+    "p": 27,
+    "r": 19,
+    "s": 25,
+    "t": 13,
+    "u": 21,
+    "w": 14,
+    "y": 16,
+    "|": 4,
+    "à": 28,
+    "â": 6,
+    "ã": 31,
+    "å": 35,
+    "æ": 11,
+    "è": 30,
+    "ì": 36,
+    "î": 34,
+    "ò": 39,
+    "ô": 15,
+    "õ": 32,
+    "ù": 33,
+    "û": 26,
+    "ü": 37,
+    "ǹ": 38
+  },
+  "adh": {
+    "'": 28,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 24,
+    "c": 23,
+    "d": 17,
+    "e": 8,
+    "f": 27,
+    "g": 19,
+    "h": 15,
+    "i": 6,
+    "j": 18,
+    "k": 10,
+    "l": 21,
+    "m": 11,
+    "n": 9,
+    "o": 5,
+    "p": 22,
+    "r": 13,
+    "s": 26,
+    "t": 16,
+    "u": 20,
+    "v": 30,
+    "w": 12,
+    "y": 14,
+    "z": 29,
+    "|": 4,
+    "ŋ": 25
+  },
+  "adj": {
+    "'": 14,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 24,
+    "d": 26,
+    "e": 6,
+    "f": 28,
+    "g": 23,
+    "h": 36,
+    "i": 12,
+    "j": 27,
+    "k": 11,
+    "l": 10,
+    "m": 9,
+    "n": 8,
+    "o": 17,
+    "p": 29,
+    "r": 22,
+    "s": 15,
+    "t": 25,
+    "u": 21,
+    "v": 43,
+    "w": 19,
+    "y": 18,
+    "z": 47,
+    "|": 4,
+    "à": 41,
+    "á": 35,
+    "è": 46,
+    "é": 37,
+    "ê": 42,
+    "ì": 38,
+    "í": 33,
+    "ò": 39,
+    "ó": 31,
+    "ô": 48,
+    "ù": 44,
+    "ú": 40,
+    "ŋ": 13,
+    "ɔ": 20,
+    "ɛ": 7,
+    "̀": 34,
+    "́": 32,
+    "̂": 45
+  },
+  "adx": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "་": 4,
+    "ཀ": 23,
+    "ཁ": 21,
+    "ག": 8,
+    "ང": 11,
+    "ཅ": 26,
+    "ཆ": 29,
+    "ཇ": 45,
+    "ཉ": 35,
+    "ཊ": 59,
+    "ཏ": 34,
+    "ཐ": 31,
+    "ད": 7,
+    "ན": 10,
+    "པ": 20,
+    "ཕ": 27,
+    "བ": 12,
+    "མ": 13,
+    "ཙ": 42,
+    "ཚ": 37,
+    "ཛ": 47,
+    "ཝ": 58,
+    "ཞ": 25,
+    "ཟ": 38,
+    "འ": 18,
+    "ཡ": 22,
+    "ར": 14,
+    "ལ": 19,
+    "ཤ": 32,
+    "ས": 5,
+    "ཧ": 51,
+    "ཨ": 49,
+    "ཪ": 60,
+    "ཱ": 39,
+    "ི": 6,
+    "ུ": 15,
+    "ེ": 16,
+    "ོ": 9,
+    "ྐ": 33,
+    "ྒ": 36,
+    "ྔ": 48,
+    "ྕ": 56,
+    "ྗ": 46,
+    "ྙ": 52,
+    "ྟ": 30,
+    "ྡ": 41,
+    "ྣ": 28,
+    "ྤ": 44,
+    "ྦ": 54,
+    "ྨ": 43,
+    "ྩ": 50,
+    "ྫ": 55,
+    "ྭ": 57,
+    "ྱ": 17,
+    "ྲ": 24,
+    "ླ": 40,
+    "ྷ": 53
+  },
+  "aeu": {
+    "-": 15,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 26,
+    "d": 21,
+    "e": 10,
+    "g": 14,
+    "h": 11,
+    "i": 7,
+    "j": 23,
+    "k": 19,
+    "l": 6,
+    "m": 18,
+    "n": 9,
+    "o": 12,
+    "p": 25,
+    "q": 8,
+    "s": 24,
+    "t": 20,
+    "u": 16,
+    "v": 13,
+    "w": 27,
+    "y": 22,
+    "|": 4
+  },
+  "afr": {
+    "\"": 68,
+    "$": 55,
+    "%": 56,
+    "&": 64,
+    "'": 26,
+    "+": 63,
+    ",": 48,
+    "-": 30,
+    ".": 38,
+    "/": 49,
+    "0": 28,
+    "1": 31,
+    "2": 34,
+    "3": 36,
+    "4": 37,
+    "5": 39,
+    "6": 40,
+    "7": 44,
+    "8": 42,
+    "9": 35,
+    ":": 47,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 65,
+    "]": 66,
+    "a": 7,
+    "b": 23,
+    "c": 29,
+    "d": 13,
+    "e": 5,
+    "f": 25,
+    "g": 15,
+    "h": 22,
+    "i": 6,
+    "j": 27,
+    "k": 16,
+    "l": 14,
+    "m": 17,
+    "n": 8,
+    "o": 11,
+    "p": 21,
+    "q": 50,
+    "r": 10,
+    "s": 9,
+    "t": 12,
+    "u": 19,
+    "v": 18,
+    "w": 20,
+    "x": 45,
+    "y": 24,
+    "z": 43,
+    "|": 4,
+    "£": 72,
+    "¥": 60,
+    "°": 73,
+    "²": 67,
+    "á": 51,
+    "ç": 70,
+    "é": 76,
+    "ê": 33,
+    "ë": 32,
+    "í": 59,
+    "ï": 46,
+    "ó": 61,
+    "ö": 69,
+    "ú": 75,
+    "ł": 71,
+    "ń": 57,
+    "ͦ": 74,
+    "ϊ": 62,
+    "–": 58,
+    "‘": 41,
+    "’": 52,
+    "”": 53
+  },
+  "agd": {
+    "'": 28,
+    "-": 29,
+    "0": 30,
+    "3": 32,
+    "4": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "d": 23,
+    "e": 9,
+    "f": 19,
+    "h": 25,
+    "i": 7,
+    "k": 13,
+    "m": 8,
+    "n": 6,
+    "o": 10,
+    "p": 24,
+    "r": 12,
+    "s": 18,
+    "t": 14,
+    "u": 15,
+    "v": 17,
+    "y": 16,
+    "|": 4,
+    "á": 11,
+    "é": 22,
+    "í": 21,
+    "ó": 27,
+    "ú": 26
+  },
+  "agg": {
+    "-": 27,
+    "0": 28,
+    "1": 29,
+    "2": 31,
+    "3": 35,
+    "4": 32,
+    "5": 33,
+    "6": 34,
+    "7": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 15,
+    "d": 13,
+    "e": 16,
+    "f": 20,
+    "g": 21,
+    "h": 10,
+    "i": 14,
+    "k": 23,
+    "m": 9,
+    "n": 8,
+    "o": 7,
+    "p": 25,
+    "r": 11,
+    "s": 17,
+    "t": 26,
+    "u": 12,
+    "w": 24,
+    "y": 19,
+    "|": 5,
+    "ü": 18,
+    "ŋ": 22,
+    "ɨ": 6
+  },
+  "agn": {
+    "'": 31,
+    "-": 22,
+    "0": 34,
+    "1": 36,
+    "2": 41,
+    "5": 40,
+    "6": 38,
+    "7": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 25,
+    "d": 12,
+    "e": 15,
+    "f": 27,
+    "g": 7,
+    "h": 26,
+    "i": 8,
+    "j": 24,
+    "k": 18,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 9,
+    "p": 16,
+    "q": 30,
+    "r": 17,
+    "s": 13,
+    "t": 10,
+    "u": 23,
+    "v": 29,
+    "w": 21,
+    "x": 35,
+    "y": 20,
+    "z": 28,
+    "|": 5,
+    "ā": 32,
+    "ē": 39,
+    "ō": 33,
+    "—": 37
+  },
+  "agr": {
+    "'": 32,
+    "2": 39,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 20,
+    "d": 18,
+    "e": 14,
+    "f": 31,
+    "g": 15,
+    "h": 16,
+    "i": 6,
+    "j": 12,
+    "k": 10,
+    "l": 28,
+    "m": 11,
+    "n": 8,
+    "o": 25,
+    "p": 19,
+    "q": 36,
+    "r": 24,
+    "s": 13,
+    "t": 9,
+    "u": 7,
+    "v": 33,
+    "w": 17,
+    "x": 37,
+    "y": 21,
+    "z": 34,
+    "|": 5,
+    "á": 26,
+    "é": 30,
+    "í": 22,
+    "ñ": 38,
+    "ó": 35,
+    "ú": 27,
+    "—": 29
+  },
+  "agu": {
+    "'": 35,
+    "-": 38,
+    "6": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 21,
+    "c": 11,
+    "d": 31,
+    "e": 10,
+    "f": 39,
+    "g": 32,
+    "h": 20,
+    "i": 7,
+    "j": 15,
+    "k": 18,
+    "l": 14,
+    "m": 24,
+    "n": 5,
+    "o": 16,
+    "p": 26,
+    "q": 27,
+    "r": 25,
+    "s": 19,
+    "t": 9,
+    "u": 12,
+    "v": 40,
+    "w": 22,
+    "x": 23,
+    "y": 13,
+    "z": 17,
+    "|": 4,
+    "á": 36,
+    "é": 34,
+    "í": 33,
+    "ñ": 41,
+    "ó": 37,
+    "ú": 29,
+    "ẍ": 28,
+    "—": 30,
+    "ꞌ": 8
+  },
+  "agx": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "а": 4,
+    "б": 26,
+    "в": 20,
+    "г": 12,
+    "д": 15,
+    "е": 9,
+    "ж": 32,
+    "з": 29,
+    "и": 6,
+    "й": 22,
+    "к": 14,
+    "л": 16,
+    "м": 25,
+    "н": 8,
+    "о": 36,
+    "п": 27,
+    "р": 10,
+    "с": 11,
+    "т": 19,
+    "у": 7,
+    "ф": 28,
+    "х": 17,
+    "ц": 34,
+    "ч": 23,
+    "ш": 30,
+    "ъ": 21,
+    "ь": 13,
+    "э": 31,
+    "ю": 35,
+    "я": 24,
+    "ӏ": 18,
+    "–": 33
+  },
+  "aha": {
+    "'": 41,
+    "0": 39,
+    "1": 43,
+    "2": 44,
+    "3": 48,
+    "4": 46,
+    "5": 45,
+    "6": 47,
+    "7": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 14,
+    "d": 20,
+    "e": 16,
+    "f": 31,
+    "g": 27,
+    "h": 25,
+    "i": 10,
+    "k": 18,
+    "l": 11,
+    "m": 9,
+    "n": 6,
+    "o": 23,
+    "p": 30,
+    "s": 21,
+    "t": 24,
+    "u": 19,
+    "v": 28,
+    "w": 17,
+    "y": 13,
+    "z": 22,
+    "|": 4,
+    "á": 32,
+    "ã": 34,
+    "é": 37,
+    "í": 36,
+    "ó": 40,
+    "ú": 42,
+    "ĩ": 33,
+    "ũ": 38,
+    "ɔ": 12,
+    "ɛ": 8,
+    "ɣ": 29,
+    "ɩ": 5,
+    "ʋ": 15,
+    "́": 35,
+    "̃": 26
+  },
+  "ahk": {
+    "'": 17,
+    "-": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 31,
+    "a": 5,
+    "b": 27,
+    "c": 30,
+    "d": 20,
+    "e": 7,
+    "f": 32,
+    "g": 19,
+    "h": 8,
+    "i": 13,
+    "j": 26,
+    "k": 22,
+    "l": 18,
+    "m": 12,
+    "n": 15,
+    "o": 24,
+    "p": 28,
+    "s": 21,
+    "t": 23,
+    "u": 9,
+    "w": 11,
+    "y": 16,
+    "z": 29,
+    "|": 4,
+    "ˆ": 25,
+    "ˬ": 6,
+    "̭": 14
+  },
+  "aia": {
+    "-": 22,
+    "6": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 31,
+    "d": 18,
+    "e": 14,
+    "f": 27,
+    "g": 17,
+    "h": 13,
+    "i": 6,
+    "j": 23,
+    "k": 21,
+    "l": 25,
+    "m": 12,
+    "n": 8,
+    "o": 9,
+    "p": 24,
+    "r": 10,
+    "s": 15,
+    "t": 16,
+    "u": 11,
+    "v": 26,
+    "w": 19,
+    "y": 29,
+    "|": 5,
+    "—": 28,
+    "ꞌ": 7
+  },
+  "aka": {
+    "'": 27,
+    "-": 29,
+    "1": 30,
+    "2": 34,
+    "3": 33,
+    "7": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 20,
+    "e": 8,
+    "f": 23,
+    "g": 25,
+    "h": 21,
+    "i": 11,
+    "k": 17,
+    "l": 26,
+    "m": 9,
+    "n": 6,
+    "o": 7,
+    "p": 24,
+    "r": 18,
+    "s": 12,
+    "t": 22,
+    "u": 16,
+    "w": 15,
+    "y": 14,
+    "|": 4,
+    "á": 32,
+    "ɔ": 13,
+    "ɛ": 10,
+    "ʼ": 28
+  },
+  "akb": {
+    "'": 25,
+    "-": 23,
+    "0": 31,
+    "1": 33,
+    "4": 32,
+    "6": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 26,
+    "d": 10,
+    "e": 17,
+    "f": 29,
+    "g": 11,
+    "h": 13,
+    "i": 7,
+    "j": 22,
+    "k": 19,
+    "l": 18,
+    "m": 14,
+    "n": 6,
+    "o": 8,
+    "p": 20,
+    "r": 16,
+    "s": 12,
+    "t": 15,
+    "u": 9,
+    "v": 30,
+    "w": 27,
+    "y": 24,
+    "z": 28,
+    "|": 4
+  },
+  "ake": {
+    "'": 6,
+    "-": 34,
+    "0": 22,
+    "1": 24,
+    "2": 26,
+    "3": 29,
+    "4": 27,
+    "5": 28,
+    "6": 30,
+    "7": 25,
+    "8": 31,
+    "9": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "e": 8,
+    "i": 11,
+    "k": 15,
+    "m": 19,
+    "n": 7,
+    "o": 12,
+    "p": 9,
+    "r": 14,
+    "s": 17,
+    "t": 13,
+    "u": 18,
+    "w": 21,
+    "y": 20,
+    "|": 4,
+    "ɨ": 10,
+    "ʉ": 16,
+    "–": 33,
+    "—": 23
+  },
+  "akp": {
+    "'": 37,
+    "-": 33,
+    "6": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "d": 27,
+    "e": 8,
+    "f": 29,
+    "g": 19,
+    "h": 34,
+    "i": 6,
+    "k": 15,
+    "l": 20,
+    "m": 9,
+    "n": 14,
+    "o": 12,
+    "p": 26,
+    "r": 13,
+    "s": 10,
+    "t": 18,
+    "u": 16,
+    "v": 35,
+    "w": 23,
+    "y": 24,
+    "z": 30,
+    "|": 4,
+    "ã": 28,
+    "ĩ": 32,
+    "ũ": 31,
+    "ɔ": 7,
+    "ɖ": 22,
+    "ɛ": 11,
+    "ɣ": 25,
+    "̃": 21
+  },
+  "alj": {
+    "0": 22,
+    "1": 24,
+    "3": 26,
+    "4": 23,
+    "5": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "d": 20,
+    "e": 16,
+    "g": 9,
+    "i": 7,
+    "k": 11,
+    "l": 19,
+    "m": 14,
+    "n": 6,
+    "o": 8,
+    "p": 15,
+    "r": 21,
+    "s": 12,
+    "t": 13,
+    "w": 18,
+    "y": 10,
+    "|": 5,
+    "ɩ": 27
+  },
+  "alp": {
+    "'": 28,
+    "-": 27,
+    "0": 29,
+    "1": 30,
+    "2": 32,
+    "4": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "c": 26,
+    "d": 22,
+    "e": 5,
+    "f": 25,
+    "g": 24,
+    "h": 19,
+    "i": 7,
+    "j": 23,
+    "k": 8,
+    "l": 9,
+    "m": 13,
+    "n": 14,
+    "o": 15,
+    "p": 18,
+    "r": 16,
+    "s": 12,
+    "t": 11,
+    "u": 10,
+    "w": 21,
+    "y": 20,
+    "|": 4
+  },
+  "alt": {
+    "-": 33,
+    "0": 48,
+    "1": 39,
+    "2": 40,
+    "4": 44,
+    "6": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "j": 46,
+    "r": 45,
+    "|": 4,
+    "ÿ": 43,
+    "а": 5,
+    "б": 18,
+    "в": 34,
+    "г": 17,
+    "д": 13,
+    "е": 6,
+    "ж": 29,
+    "з": 27,
+    "и": 11,
+    "й": 20,
+    "к": 12,
+    "л": 10,
+    "м": 25,
+    "н": 9,
+    "о": 14,
+    "п": 21,
+    "р": 7,
+    "с": 19,
+    "т": 15,
+    "у": 16,
+    "ф": 36,
+    "х": 35,
+    "ц": 41,
+    "ч": 26,
+    "ш": 30,
+    "ы": 8,
+    "ь": 38,
+    "э": 31,
+    "ю": 42,
+    "я": 37,
+    "ј": 22,
+    "ҥ": 24,
+    "ӧ": 28,
+    "ӱ": 23,
+    "–": 32
+  },
+  "alz": {
+    "'": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 19,
+    "d": 14,
+    "e": 7,
+    "f": 27,
+    "g": 13,
+    "h": 23,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 21,
+    "m": 11,
+    "n": 8,
+    "o": 10,
+    "p": 24,
+    "r": 16,
+    "s": 26,
+    "t": 17,
+    "u": 9,
+    "v": 28,
+    "w": 15,
+    "y": 20,
+    "z": 29,
+    "|": 4
+  },
+  "ame": {
+    "'": 11,
+    "-": 40,
+    "2": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 18,
+    "d": 27,
+    "e": 6,
+    "f": 34,
+    "g": 31,
+    "h": 9,
+    "i": 23,
+    "j": 24,
+    "k": 20,
+    "l": 17,
+    "m": 16,
+    "n": 14,
+    "o": 8,
+    "p": 13,
+    "q": 39,
+    "r": 12,
+    "s": 15,
+    "t": 7,
+    "u": 26,
+    "v": 35,
+    "w": 21,
+    "x": 22,
+    "y": 19,
+    "z": 38,
+    "|": 5,
+    "á": 36,
+    "é": 32,
+    "ë": 30,
+    "í": 33,
+    "ñ": 10,
+    "ó": 37,
+    "ú": 28,
+    "—": 29
+  },
+  "amf": {
+    "'": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 11,
+    "e": 7,
+    "g": 19,
+    "h": 13,
+    "i": 8,
+    "j": 25,
+    "k": 12,
+    "l": 21,
+    "m": 14,
+    "n": 6,
+    "o": 9,
+    "p": 23,
+    "q": 24,
+    "r": 17,
+    "s": 10,
+    "t": 16,
+    "u": 22,
+    "v": 28,
+    "w": 20,
+    "x": 30,
+    "y": 15,
+    "z": 27,
+    "|": 5
+  },
+  "amh": {
+    "!": 297,
+    "\"": 242,
+    "$": 256,
+    "%": 264,
+    "&": 303,
+    "'": 293,
+    "+": 276,
+    ",": 194,
+    "-": 142,
+    ".": 138,
+    "/": 208,
+    "0": 64,
+    "1": 77,
+    "2": 104,
+    "3": 153,
+    "4": 149,
+    "5": 144,
+    "6": 157,
+    "7": 180,
+    "8": 159,
+    "9": 139,
+    ":": 243,
+    ";": 286,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 273,
+    "[": 306,
+    "]": 307,
+    "a": 113,
+    "b": 205,
+    "c": 173,
+    "d": 203,
+    "e": 146,
+    "f": 218,
+    "g": 197,
+    "h": 186,
+    "i": 152,
+    "j": 238,
+    "k": 217,
+    "l": 165,
+    "m": 182,
+    "n": 141,
+    "o": 135,
+    "p": 183,
+    "q": 302,
+    "r": 158,
+    "s": 164,
+    "t": 160,
+    "u": 175,
+    "v": 227,
+    "w": 226,
+    "x": 255,
+    "y": 221,
+    "z": 266,
+    "|": 4,
+    "£": 304,
+    "¥": 288,
+    "«": 260,
+    "°": 279,
+    "²": 298,
+    "»": 261,
+    "ã": 287,
+    "õ": 283,
+    "ሀ": 123,
+    "ሁ": 63,
+    "ሂ": 184,
+    "ሃ": 95,
+    "ሄ": 137,
+    "ህ": 51,
+    "ሆ": 48,
+    "ለ": 18,
+    "ሉ": 43,
+    "ሊ": 61,
+    "ላ": 24,
+    "ሌ": 82,
+    "ል": 13,
+    "ሎ": 62,
+    "ሏ": 185,
+    "ሐ": 189,
+    "ሑ": 235,
+    "ሒ": 338,
+    "ሓ": 313,
+    "ሔ": 233,
+    "ሕ": 130,
+    "ሖ": 331,
+    "መ": 12,
+    "ሙ": 117,
+    "ሚ": 38,
+    "ማ": 29,
+    "ሜ": 89,
+    "ም": 14,
+    "ሞ": 79,
+    "ሟ": 212,
+    "ሠ": 174,
+    "ሡ": 316,
+    "ሢ": 336,
+    "ሣ": 228,
+    "ሤ": 321,
+    "ሥ": 132,
+    "ሦ": 229,
+    "ረ": 30,
+    "ሩ": 65,
+    "ሪ": 44,
+    "ራ": 35,
+    "ሬ": 111,
+    "ር": 11,
+    "ሮ": 58,
+    "ሯ": 196,
+    "ሰ": 36,
+    "ሱ": 112,
+    "ሲ": 70,
+    "ሳ": 47,
+    "ሴ": 134,
+    "ስ": 10,
+    "ሶ": 119,
+    "ሷ": 223,
+    "ሸ": 145,
+    "ሹ": 207,
+    "ሺ": 192,
+    "ሻ": 121,
+    "ሼ": 275,
+    "ሽ": 86,
+    "ሾ": 209,
+    "ሿ": 280,
+    "ቀ": 42,
+    "ቁ": 105,
+    "ቂ": 147,
+    "ቃ": 67,
+    "ቄ": 219,
+    "ቅ": 50,
+    "ቆ": 107,
+    "ቈ": 310,
+    "ቊ": 315,
+    "ቋ": 116,
+    "ቍ": 314,
+    "በ": 6,
+    "ቡ": 92,
+    "ቢ": 72,
+    "ባ": 40,
+    "ቤ": 97,
+    "ብ": 28,
+    "ቦ": 80,
+    "ቧ": 210,
+    "ቨ": 206,
+    "ቪ": 168,
+    "ቫ": 190,
+    "ቬ": 211,
+    "ቭ": 213,
+    "ቮ": 259,
+    "ተ": 19,
+    "ቱ": 66,
+    "ቲ": 114,
+    "ታ": 34,
+    "ቴ": 129,
+    "ት": 7,
+    "ቶ": 56,
+    "ቷ": 169,
+    "ቸ": 55,
+    "ቹ": 103,
+    "ቺ": 220,
+    "ቻ": 98,
+    "ቼ": 241,
+    "ች": 15,
+    "ቾ": 232,
+    "ቿ": 299,
+    "ኀ": 309,
+    "ኃ": 172,
+    "ኄ": 328,
+    "ኅ": 251,
+    "ኆ": 330,
+    "ኋ": 150,
+    "ነ": 21,
+    "ኑ": 83,
+    "ኒ": 99,
+    "ና": 23,
+    "ኔ": 118,
+    "ን": 5,
+    "ኖ": 69,
+    "ኗ": 237,
+    "ኘ": 154,
+    "ኙ": 163,
+    "ኚ": 244,
+    "ኛ": 57,
+    "ኜ": 292,
+    "ኝ": 124,
+    "ኞ": 179,
+    "ኟ": 305,
+    "አ": 17,
+    "ኡ": 257,
+    "ኢ": 126,
+    "ኣ": 252,
+    "ኤ": 115,
+    "እ": 20,
+    "ኦ": 143,
+    "ከ": 25,
+    "ኩ": 136,
+    "ኪ": 110,
+    "ካ": 46,
+    "ኬ": 176,
+    "ክ": 39,
+    "ኮ": 88,
+    "ኰ": 300,
+    "ኳ": 181,
+    "ኵ": 312,
+    "ኸ": 285,
+    "ኹ": 319,
+    "ኺ": 289,
+    "ኻ": 324,
+    "ኽ": 323,
+    "ኾ": 320,
+    "ወ": 33,
+    "ዉ": 250,
+    "ዊ": 74,
+    "ዋ": 53,
+    "ዌ": 200,
+    "ው": 9,
+    "ዎ": 41,
+    "ዐ": 240,
+    "ዑ": 249,
+    "ዒ": 278,
+    "ዓ": 90,
+    "ዔ": 325,
+    "ዕ": 101,
+    "ዖ": 281,
+    "ዘ": 71,
+    "ዙ": 87,
+    "ዚ": 91,
+    "ዛ": 84,
+    "ዜ": 81,
+    "ዝ": 68,
+    "ዞ": 167,
+    "ዟ": 265,
+    "ዠ": 332,
+    "ዡ": 267,
+    "ዢ": 248,
+    "ዣ": 231,
+    "ዤ": 329,
+    "ዥ": 201,
+    "ዦ": 245,
+    "የ": 8,
+    "ዩ": 100,
+    "ዪ": 272,
+    "ያ": 22,
+    "ዬ": 225,
+    "ይ": 16,
+    "ዮ": 128,
+    "ደ": 26,
+    "ዱ": 108,
+    "ዲ": 75,
+    "ዳ": 52,
+    "ዴ": 177,
+    "ድ": 27,
+    "ዶ": 94,
+    "ዷ": 246,
+    "ጀ": 85,
+    "ጁ": 202,
+    "ጂ": 166,
+    "ጃ": 133,
+    "ጄ": 224,
+    "ጅ": 109,
+    "ጆ": 171,
+    "ጇ": 294,
+    "ገ": 32,
+    "ጉ": 78,
+    "ጊ": 76,
+    "ጋ": 54,
+    "ጌ": 230,
+    "ግ": 31,
+    "ጎ": 96,
+    "ጐ": 282,
+    "ጒ": 311,
+    "ጓ": 140,
+    "ጕ": 322,
+    "ጠ": 45,
+    "ጡ": 151,
+    "ጢ": 258,
+    "ጣ": 59,
+    "ጤ": 193,
+    "ጥ": 37,
+    "ጦ": 162,
+    "ጧ": 236,
+    "ጨ": 102,
+    "ጩ": 234,
+    "ጪ": 216,
+    "ጫ": 127,
+    "ጬ": 326,
+    "ጭ": 131,
+    "ጮ": 239,
+    "ጯ": 333,
+    "ጰ": 335,
+    "ጱ": 334,
+    "ጲ": 317,
+    "ጳ": 291,
+    "ጴ": 290,
+    "ጵ": 268,
+    "ጶ": 318,
+    "ጸ": 191,
+    "ጹ": 263,
+    "ጺ": 295,
+    "ጻ": 199,
+    "ጼ": 327,
+    "ጽ": 156,
+    "ጾ": 254,
+    "ጿ": 271,
+    "ፀ": 187,
+    "ፁ": 270,
+    "ፃ": 178,
+    "ፄ": 337,
+    "ፅ": 170,
+    "ፆ": 253,
+    "ፈ": 60,
+    "ፉ": 161,
+    "ፊ": 93,
+    "ፋ": 106,
+    "ፌ": 195,
+    "ፍ": 49,
+    "ፎ": 148,
+    "ፏ": 222,
+    "ፐ": 204,
+    "ፑ": 262,
+    "ፒ": 155,
+    "ፓ": 120,
+    "ፔ": 215,
+    "ፕ": 122,
+    "ፖ": 125,
+    "ፗ": 301,
+    "፡": 198,
+    "።": 188,
+    "፣": 73,
+    "፤": 247,
+    "፥": 277,
+    "፦": 308,
+    "–": 284,
+    "—": 269,
+    "‘": 296,
+    "’": 274,
+    "”": 214
+  },
+  "ami": {
+    "'": 21,
+    "-": 28,
+    "0": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 17,
+    "d": 20,
+    "e": 23,
+    "f": 24,
+    "g": 19,
+    "h": 22,
+    "i": 8,
+    "k": 9,
+    "l": 15,
+    "m": 11,
+    "n": 7,
+    "o": 6,
+    "p": 18,
+    "r": 14,
+    "s": 13,
+    "t": 10,
+    "u": 25,
+    "w": 16,
+    "x": 26,
+    "y": 12,
+    "|": 5
+  },
+  "amk": {
+    "'": 29,
+    "-": 26,
+    "0": 27,
+    "1": 30,
+    "2": 31,
+    "3": 37,
+    "4": 33,
+    "6": 34,
+    "7": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 32,
+    "d": 18,
+    "e": 8,
+    "f": 15,
+    "g": 19,
+    "h": 20,
+    "i": 6,
+    "j": 25,
+    "k": 17,
+    "l": 24,
+    "m": 11,
+    "n": 7,
+    "o": 9,
+    "p": 23,
+    "r": 10,
+    "s": 16,
+    "t": 12,
+    "u": 13,
+    "w": 14,
+    "y": 22,
+    "z": 28,
+    "|": 4,
+    "—": 35
+  },
+  "ann": {
+    "'": 41,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 13,
+    "c": 28,
+    "d": 32,
+    "e": 5,
+    "f": 29,
+    "g": 18,
+    "h": 27,
+    "i": 6,
+    "j": 22,
+    "k": 9,
+    "l": 23,
+    "m": 11,
+    "n": 8,
+    "o": 14,
+    "p": 24,
+    "r": 19,
+    "s": 21,
+    "t": 15,
+    "u": 10,
+    "w": 20,
+    "y": 17,
+    "|": 4,
+    "à": 39,
+    "â": 37,
+    "è": 25,
+    "ê": 36,
+    "ì": 26,
+    "î": 31,
+    "ò": 30,
+    "ô": 34,
+    "ù": 40,
+    "û": 42,
+    "ǹ": 35,
+    "̀": 38,
+    "̄": 16,
+    "ọ": 12
+  },
+  "any": {
+    "'": 15,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 14,
+    "c": 34,
+    "d": 25,
+    "e": 21,
+    "f": 24,
+    "g": 29,
+    "h": 27,
+    "i": 16,
+    "j": 32,
+    "k": 13,
+    "l": 11,
+    "m": 10,
+    "n": 5,
+    "o": 28,
+    "p": 30,
+    "s": 18,
+    "t": 23,
+    "u": 20,
+    "v": 37,
+    "w": 22,
+    "y": 17,
+    "z": 26,
+    "|": 4,
+    "á": 33,
+    "â": 36,
+    "é": 40,
+    "í": 39,
+    "ó": 41,
+    "ú": 38,
+    "ɔ": 8,
+    "ɛ": 7,
+    "ɩ": 9,
+    "ʋ": 12,
+    "́": 19,
+    "̂": 35
+  },
+  "aoz": {
+    "'": 14,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 26,
+    "e": 7,
+    "f": 18,
+    "g": 22,
+    "h": 17,
+    "i": 8,
+    "j": 24,
+    "k": 16,
+    "l": 15,
+    "m": 12,
+    "n": 6,
+    "o": 11,
+    "p": 20,
+    "r": 25,
+    "s": 10,
+    "t": 13,
+    "u": 9,
+    "w": 27,
+    "y": 21,
+    "|": 4
+  },
+  "apb": {
+    "'": 39,
+    "-": 23,
+    "0": 36,
+    "1": 26,
+    "2": 28,
+    "3": 29,
+    "4": 32,
+    "5": 31,
+    "6": 33,
+    "7": 37,
+    "8": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "d": 21,
+    "e": 6,
+    "f": 30,
+    "g": 18,
+    "h": 15,
+    "i": 7,
+    "j": 24,
+    "k": 14,
+    "l": 11,
+    "m": 12,
+    "n": 9,
+    "o": 8,
+    "p": 22,
+    "r": 17,
+    "s": 19,
+    "t": 16,
+    "u": 10,
+    "v": 27,
+    "w": 20,
+    "z": 40,
+    "|": 4,
+    "—": 38,
+    "ꞌ": 13
+  },
+  "apr": {
+    "'": 32,
+    "0": 23,
+    "1": 24,
+    "2": 28,
+    "3": 31,
+    "4": 25,
+    "5": 26,
+    "6": 27,
+    "7": 29,
+    "8": 33,
+    "9": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 17,
+    "e": 9,
+    "g": 12,
+    "i": 8,
+    "k": 13,
+    "l": 11,
+    "m": 14,
+    "n": 6,
+    "o": 7,
+    "p": 18,
+    "r": 20,
+    "s": 21,
+    "t": 10,
+    "u": 15,
+    "w": 22,
+    "y": 16,
+    "|": 4
+  },
+  "ara": {
+    "\"": 63,
+    "%": 83,
+    "'": 116,
+    "+": 97,
+    ",": 76,
+    "-": 59,
+    ".": 60,
+    "/": 62,
+    "0": 42,
+    "1": 43,
+    "2": 48,
+    "3": 53,
+    "4": 51,
+    "5": 52,
+    "6": 54,
+    "7": 58,
+    "8": 56,
+    "9": 50,
+    ":": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 61,
+    "b": 82,
+    "c": 74,
+    "d": 73,
+    "e": 66,
+    "f": 94,
+    "g": 72,
+    "h": 85,
+    "i": 77,
+    "j": 99,
+    "k": 90,
+    "l": 75,
+    "m": 79,
+    "n": 64,
+    "o": 65,
+    "p": 71,
+    "r": 80,
+    "s": 68,
+    "t": 69,
+    "u": 84,
+    "v": 81,
+    "w": 88,
+    "x": 93,
+    "y": 101,
+    "z": 96,
+    "|": 4,
+    "×": 106,
+    "õ": 98,
+    "،": 46,
+    "؛": 86,
+    "ء": 37,
+    "آ": 45,
+    "أ": 20,
+    "ؤ": 47,
+    "إ": 30,
+    "ئ": 38,
+    "ا": 5,
+    "ب": 13,
+    "ة": 14,
+    "ت": 11,
+    "ث": 31,
+    "ج": 24,
+    "ح": 23,
+    "خ": 29,
+    "د": 16,
+    "ذ": 33,
+    "ر": 12,
+    "ز": 35,
+    "س": 18,
+    "ش": 27,
+    "ص": 28,
+    "ض": 32,
+    "ط": 25,
+    "ظ": 40,
+    "ع": 15,
+    "غ": 36,
+    "ـ": 91,
+    "ف": 17,
+    "ق": 21,
+    "ك": 19,
+    "ل": 6,
+    "م": 8,
+    "ن": 9,
+    "ه": 22,
+    "و": 10,
+    "ى": 26,
+    "ي": 7,
+    "ً": 34,
+    "ٌ": 57,
+    "ٍ": 44,
+    "َ": 55,
+    "ُ": 41,
+    "ِ": 49,
+    "ّ": 39,
+    "٠": 70,
+    "١": 78,
+    "٢": 103,
+    "٣": 92,
+    "٥": 107,
+    "٦": 104,
+    "٧": 100,
+    "٨": 95,
+    "٩": 87,
+    "٪": 105,
+    "ٰ": 108,
+    "چ": 102,
+    "ڨ": 117,
+    "ک": 111,
+    "ھ": 115,
+    "ی": 109,
+    "ۖ": 110,
+    "ۗ": 113,
+    "ۘ": 118,
+    "ۚ": 112,
+    "ۛ": 114,
+    "–": 120,
+    "—": 89,
+    "☭": 119
+  },
+  "arl": {
+    "'": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 13,
+    "d": 22,
+    "e": 12,
+    "f": 27,
+    "g": 26,
+    "h": 19,
+    "i": 5,
+    "j": 9,
+    "k": 36,
+    "l": 21,
+    "m": 20,
+    "n": 8,
+    "o": 11,
+    "p": 18,
+    "q": 16,
+    "r": 10,
+    "s": 14,
+    "t": 15,
+    "u": 7,
+    "v": 31,
+    "x": 34,
+    "y": 17,
+    "z": 32,
+    "|": 6,
+    "á": 29,
+    "é": 25,
+    "í": 28,
+    "ó": 30,
+    "ú": 35,
+    "—": 24
+  },
+  "asa": {
+    "'": 29,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 26,
+    "d": 22,
+    "e": 7,
+    "f": 27,
+    "g": 20,
+    "h": 14,
+    "i": 6,
+    "j": 24,
+    "k": 10,
+    "l": 21,
+    "m": 11,
+    "n": 8,
+    "o": 12,
+    "p": 28,
+    "r": 17,
+    "s": 18,
+    "t": 16,
+    "u": 9,
+    "v": 13,
+    "w": 15,
+    "y": 19,
+    "z": 25,
+    "|": 4
+  },
+  "asg": {
+    "'": 34,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 26,
+    "d": 22,
+    "e": 11,
+    "f": 30,
+    "g": 20,
+    "h": 23,
+    "i": 6,
+    "j": 32,
+    "k": 9,
+    "l": 17,
+    "m": 12,
+    "n": 8,
+    "o": 15,
+    "p": 25,
+    "r": 29,
+    "s": 13,
+    "t": 14,
+    "u": 7,
+    "v": 21,
+    "w": 24,
+    "y": 16,
+    "z": 28,
+    "|": 4,
+    "ɓ": 31,
+    "ɗ": 18,
+    "̱": 10,
+    "ꞌ": 27
+  },
+  "asm": {
+    "!": 124,
+    "\"": 51,
+    "$": 112,
+    "%": 117,
+    "&": 132,
+    "'": 47,
+    "+": 136,
+    ",": 75,
+    "-": 62,
+    ".": 77,
+    "/": 100,
+    "0": 54,
+    "1": 58,
+    "2": 61,
+    "3": 67,
+    "4": 70,
+    "5": 66,
+    "6": 72,
+    "7": 76,
+    "8": 71,
+    "9": 64,
+    ":": 90,
+    ";": 116,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 74,
+    "b": 103,
+    "c": 84,
+    "d": 102,
+    "e": 83,
+    "f": 115,
+    "g": 101,
+    "h": 105,
+    "i": 85,
+    "j": 123,
+    "k": 118,
+    "l": 96,
+    "m": 91,
+    "n": 88,
+    "o": 92,
+    "p": 89,
+    "q": 131,
+    "r": 98,
+    "s": 80,
+    "t": 95,
+    "u": 87,
+    "v": 110,
+    "w": 120,
+    "x": 125,
+    "y": 121,
+    "z": 119,
+    "|": 4,
+    "£": 135,
+    "°": 130,
+    "²": 134,
+    "õ": 129,
+    "ʼ": 99,
+    "̇": 133,
+    "।": 53,
+    "ঁ": 49,
+    "ং": 48,
+    "ঃ": 79,
+    "অ": 39,
+    "আ": 25,
+    "ই": 28,
+    "ঈ": 127,
+    "উ": 46,
+    "ঋ": 128,
+    "এ": 38,
+    "ঐ": 109,
+    "ও": 43,
+    "ঔ": 114,
+    "ক": 9,
+    "খ": 42,
+    "গ": 29,
+    "ঘ": 56,
+    "ঙ": 63,
+    "চ": 32,
+    "ছ": 24,
+    "জ": 30,
+    "ঝ": 113,
+    "ঞ": 59,
+    "ট": 23,
+    "ঠ": 55,
+    "ড": 44,
+    "ঢ": 68,
+    "ণ": 37,
+    "ত": 11,
+    "থ": 40,
+    "দ": 26,
+    "ধ": 41,
+    "ন": 12,
+    "প": 16,
+    "ফ": 50,
+    "ব": 13,
+    "ভ": 36,
+    "ম": 18,
+    "য": 14,
+    "র": 57,
+    "ল": 15,
+    "শ": 35,
+    "ষ": 34,
+    "স": 17,
+    "হ": 20,
+    "়": 22,
+    "া": 6,
+    "ি": 7,
+    "ী": 27,
+    "ু": 19,
+    "ূ": 45,
+    "ৃ": 52,
+    "ে": 10,
+    "ৈ": 33,
+    "ো": 21,
+    "ৌ": 69,
+    "্": 8,
+    "ৎ": 65,
+    "০": 73,
+    "১": 78,
+    "২": 81,
+    "৩": 97,
+    "৪": 82,
+    "৫": 94,
+    "৬": 93,
+    "৭": 108,
+    "৮": 104,
+    "৯": 106,
+    "ৰ": 5,
+    "ৱ": 31,
+    "৷": 60,
+    "‌": 107,
+    "‍": 137,
+    "–": 126,
+    "—": 122,
+    "’": 86,
+    "”": 111
+  },
+  "ast": {
+    "!": 59,
+    "$": 65,
+    "%": 63,
+    "&": 68,
+    "'": 33,
+    "+": 73,
+    ",": 54,
+    "-": 43,
+    ".": 40,
+    "/": 55,
+    "0": 34,
+    "1": 36,
+    "2": 39,
+    "3": 46,
+    "4": 44,
+    "5": 45,
+    "6": 48,
+    "7": 51,
+    "8": 47,
+    "9": 41,
+    ":": 53,
+    ";": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 61,
+    "[": 76,
+    "]": 77,
+    "a": 6,
+    "b": 19,
+    "c": 16,
+    "d": 15,
+    "e": 5,
+    "f": 22,
+    "g": 21,
+    "h": 28,
+    "i": 11,
+    "j": 49,
+    "k": 38,
+    "l": 12,
+    "m": 17,
+    "n": 8,
+    "o": 9,
+    "p": 18,
+    "q": 24,
+    "r": 10,
+    "s": 7,
+    "t": 14,
+    "u": 13,
+    "v": 23,
+    "w": 42,
+    "x": 27,
+    "y": 20,
+    "z": 32,
+    "|": 4,
+    "¡": 62,
+    "¥": 70,
+    "ª": 72,
+    "²": 74,
+    "º": 64,
+    "¿": 60,
+    "á": 25,
+    "ç": 69,
+    "é": 31,
+    "í": 29,
+    "ñ": 35,
+    "ó": 26,
+    "õ": 67,
+    "ö": 78,
+    "ú": 37,
+    "ü": 52,
+    "ý": 75,
+    "ł": 79,
+    "ś": 71,
+    "ḥ": 58,
+    "–": 57,
+    "—": 66,
+    "’": 30,
+    "”": 50
+  },
+  "ata": {
+    "'": 16,
+    "-": 25,
+    "0": 24,
+    "1": 28,
+    "2": 29,
+    "4": 27,
+    "5": 34,
+    "6": 33,
+    "7": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "d": 21,
+    "e": 11,
+    "f": 26,
+    "g": 18,
+    "h": 31,
+    "i": 6,
+    "k": 19,
+    "l": 10,
+    "m": 12,
+    "n": 9,
+    "o": 8,
+    "p": 20,
+    "r": 22,
+    "s": 15,
+    "t": 14,
+    "u": 7,
+    "v": 17,
+    "x": 13,
+    "z": 32,
+    "|": 4,
+    "–": 30
+  },
+  "atb": {
+    "'": 44,
+    "-": 37,
+    "0": 46,
+    "1": 47,
+    "2": 48,
+    "4": 50,
+    "6": 51,
+    "8": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 23,
+    "c": 39,
+    "d": 29,
+    "e": 22,
+    "g": 5,
+    "h": 7,
+    "i": 12,
+    "j": 35,
+    "k": 16,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 18,
+    "p": 28,
+    "q": 17,
+    "r": 20,
+    "s": 19,
+    "t": 15,
+    "u": 9,
+    "w": 31,
+    "x": 45,
+    "y": 10,
+    "z": 36,
+    "|": 4,
+    "à": 42,
+    "á": 25,
+    "â": 21,
+    "è": 43,
+    "é": 13,
+    "ê": 30,
+    "ì": 41,
+    "í": 38,
+    "î": 27,
+    "ò": 40,
+    "ó": 33,
+    "ô": 24,
+    "ù": 34,
+    "ú": 32,
+    "û": 26,
+    "‐": 49
+  },
+  "atg": {
+    "'": 31,
+    "-": 27,
+    "0": 35,
+    "1": 48,
+    "2": 42,
+    "3": 52,
+    "4": 45,
+    "6": 51,
+    "7": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 18,
+    "c": 32,
+    "d": 24,
+    "e": 7,
+    "f": 26,
+    "g": 14,
+    "h": 10,
+    "i": 5,
+    "j": 28,
+    "k": 17,
+    "l": 13,
+    "m": 16,
+    "n": 11,
+    "o": 12,
+    "p": 23,
+    "r": 21,
+    "s": 19,
+    "t": 22,
+    "u": 15,
+    "v": 30,
+    "w": 25,
+    "y": 20,
+    "z": 29,
+    "|": 4,
+    "à": 36,
+    "á": 33,
+    "è": 49,
+    "é": 39,
+    "ò": 46,
+    "ó": 37,
+    "ù": 50,
+    "ú": 41,
+    "ā": 40,
+    "̀": 38,
+    "́": 34,
+    "̄": 43,
+    "̌": 44,
+    "ẹ": 8,
+    "ọ": 9
+  },
+  "ati": {
+    "'": 5,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 12,
+    "c": 33,
+    "d": 27,
+    "e": 11,
+    "f": 26,
+    "g": 29,
+    "h": 15,
+    "i": 9,
+    "j": 31,
+    "k": 10,
+    "l": 18,
+    "m": 16,
+    "n": 6,
+    "o": 22,
+    "p": 23,
+    "r": 30,
+    "s": 17,
+    "t": 25,
+    "u": 13,
+    "v": 32,
+    "w": 28,
+    "y": 21,
+    "z": 20,
+    "|": 4,
+    "ë": 24,
+    "ö": 19,
+    "ɔ": 14,
+    "ɛ": 8
+  },
+  "atq": {
+    "'": 14,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 30,
+    "d": 20,
+    "e": 11,
+    "f": 26,
+    "g": 10,
+    "h": 18,
+    "i": 7,
+    "j": 28,
+    "k": 16,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 8,
+    "p": 17,
+    "r": 24,
+    "s": 19,
+    "t": 12,
+    "u": 9,
+    "w": 29,
+    "y": 22,
+    "z": 27,
+    "|": 5,
+    "ä": 23
+  },
+  "ava": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "а": 4,
+    "б": 13,
+    "в": 21,
+    "г": 8,
+    "д": 16,
+    "е": 10,
+    "ж": 28,
+    "з": 23,
+    "и": 7,
+    "й": 31,
+    "к": 18,
+    "л": 6,
+    "м": 24,
+    "н": 11,
+    "о": 19,
+    "п": 32,
+    "р": 14,
+    "с": 22,
+    "т": 25,
+    "у": 9,
+    "ф": 37,
+    "х": 20,
+    "ц": 26,
+    "ч": 27,
+    "ш": 35,
+    "щ": 30,
+    "ъ": 15,
+    "ь": 17,
+    "э": 36,
+    "ю": 39,
+    "я": 29,
+    "ё": 38,
+    "ӏ": 12,
+    "–": 33
+  },
+  "avn": {
+    "'": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 51,
+    "d": 22,
+    "e": 8,
+    "f": 30,
+    "g": 20,
+    "h": 33,
+    "i": 6,
+    "k": 11,
+    "l": 9,
+    "m": 16,
+    "n": 12,
+    "o": 18,
+    "p": 21,
+    "r": 27,
+    "s": 15,
+    "t": 19,
+    "u": 14,
+    "v": 36,
+    "w": 24,
+    "x": 29,
+    "y": 17,
+    "z": 23,
+    "|": 4,
+    "à": 43,
+    "á": 35,
+    "ã": 32,
+    "è": 41,
+    "é": 39,
+    "ì": 37,
+    "í": 25,
+    "ò": 46,
+    "ó": 44,
+    "ù": 40,
+    "ú": 38,
+    "ĩ": 48,
+    "ŋ": 28,
+    "ũ": 45,
+    "ƒ": 50,
+    "ɔ": 10,
+    "ɖ": 42,
+    "ɛ": 7,
+    "ʋ": 26,
+    "̀": 31,
+    "́": 34,
+    "̃": 49
+  },
+  "avu": {
+    "'": 47,
+    "-": 40,
+    "0": 57,
+    "1": 48,
+    "2": 49,
+    "3": 52,
+    "4": 51,
+    "5": 53,
+    "6": 55,
+    "7": 58,
+    "8": 54,
+    "9": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "b": 21,
+    "c": 43,
+    "d": 11,
+    "f": 41,
+    "g": 19,
+    "h": 50,
+    "i": 59,
+    "j": 32,
+    "k": 22,
+    "l": 9,
+    "m": 27,
+    "n": 15,
+    "p": 38,
+    "r": 5,
+    "s": 26,
+    "t": 16,
+    "v": 33,
+    "w": 45,
+    "y": 20,
+    "z": 29,
+    "|": 4,
+    "á": 7,
+    "â": 10,
+    "ã": 8,
+    "é": 24,
+    "ê": 34,
+    "í": 14,
+    "î": 12,
+    "ó": 17,
+    "ô": 25,
+    "õ": 18,
+    "ú": 36,
+    "û": 42,
+    "ĩ": 13,
+    "ũ": 37,
+    "ʼ": 6,
+    "́": 30,
+    "̂": 39,
+    "̃": 28,
+    "ṇ": 46,
+    "ạ": 35,
+    "ậ": 44,
+    "ẽ": 31,
+    "ị": 23
+  },
+  "awa": {
+    "'": 63,
+    "-": 56,
+    "0": 64,
+    "1": 68,
+    "2": 69,
+    "3": 79,
+    "4": 73,
+    "5": 80,
+    "6": 72,
+    "7": 76,
+    "9": 74,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "i": 78,
+    "|": 4,
+    "ँ": 25,
+    "ं": 34,
+    "ः": 61,
+    "अ": 21,
+    "आ": 35,
+    "इ": 15,
+    "ई": 39,
+    "उ": 16,
+    "ऊ": 54,
+    "ऋ": 71,
+    "ऍ": 55,
+    "ऎ": 75,
+    "ए": 38,
+    "ऐ": 66,
+    "ओ": 33,
+    "औ": 70,
+    "क": 6,
+    "ख": 37,
+    "ग": 31,
+    "घ": 50,
+    "च": 32,
+    "छ": 42,
+    "ज": 22,
+    "झ": 52,
+    "ञ": 60,
+    "ट": 40,
+    "ठ": 47,
+    "ड": 43,
+    "ढ": 51,
+    "ण": 48,
+    "त": 12,
+    "थ": 46,
+    "द": 27,
+    "ध": 41,
+    "न": 9,
+    "ऩ": 58,
+    "प": 18,
+    "फ": 45,
+    "ब": 17,
+    "भ": 36,
+    "म": 13,
+    "य": 28,
+    "र": 7,
+    "ऱ": 77,
+    "ल": 23,
+    "व": 26,
+    "श": 67,
+    "ष": 65,
+    "स": 10,
+    "ह": 8,
+    "़": 44,
+    "ा": 5,
+    "ि": 14,
+    "ी": 20,
+    "ु": 30,
+    "ू": 29,
+    "ृ": 57,
+    "ॅ": 59,
+    "े": 11,
+    "ै": 53,
+    "ॉ": 62,
+    "ो": 24,
+    "ौ": 49,
+    "्": 19
+  },
+  "awb": {
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 33,
+    "4": 31,
+    "5": 32,
+    "6": 35,
+    "7": 34,
+    "8": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "e": 6,
+    "g": 23,
+    "h": 9,
+    "i": 10,
+    "k": 16,
+    "m": 15,
+    "n": 7,
+    "o": 20,
+    "p": 24,
+    "q": 19,
+    "r": 8,
+    "s": 18,
+    "t": 17,
+    "u": 22,
+    "w": 12,
+    "y": 27,
+    "|": 4,
+    "á": 11,
+    "é": 13,
+    "í": 14,
+    "ó": 25,
+    "ú": 26
+  },
+  "ayo": {
+    "'": 49,
+    "-": 48,
+    "0": 42,
+    "1": 47,
+    "2": 45,
+    "3": 53,
+    "4": 46,
+    "5": 51,
+    "6": 52,
+    "7": 54,
+    "8": 58,
+    "9": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "c": 15,
+    "d": 14,
+    "e": 5,
+    "f": 39,
+    "g": 13,
+    "h": 22,
+    "i": 8,
+    "j": 10,
+    "k": 55,
+    "l": 33,
+    "m": 21,
+    "n": 11,
+    "o": 9,
+    "p": 20,
+    "q": 19,
+    "r": 12,
+    "s": 17,
+    "t": 16,
+    "u": 7,
+    "v": 41,
+    "x": 50,
+    "y": 23,
+    "z": 43,
+    "|": 4,
+    "­": 57,
+    "á": 24,
+    "ã": 32,
+    "é": 18,
+    "í": 26,
+    "ñ": 29,
+    "ó": 27,
+    "õ": 34,
+    "ú": 28,
+    "ĩ": 30,
+    "ı": 40,
+    "ũ": 38,
+    "́": 35,
+    "̃": 44,
+    "ṍ": 36,
+    "ṹ": 59,
+    "ẽ": 37,
+    "–": 31
+  },
+  "ayr": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 21,
+    "d": 28,
+    "e": 27,
+    "f": 33,
+    "g": 32,
+    "h": 17,
+    "i": 6,
+    "j": 15,
+    "k": 8,
+    "l": 22,
+    "m": 14,
+    "n": 9,
+    "o": 26,
+    "p": 12,
+    "q": 20,
+    "r": 13,
+    "s": 11,
+    "t": 7,
+    "u": 10,
+    "v": 35,
+    "w": 18,
+    "x": 16,
+    "y": 19,
+    "z": 39,
+    "|": 5,
+    "á": 40,
+    "ä": 23,
+    "é": 37,
+    "ë": 43,
+    "í": 36,
+    "ï": 29,
+    "ñ": 24,
+    "ó": 38,
+    "ö": 41,
+    "ú": 42,
+    "ü": 34,
+    "—": 30
+  },
+  "ayz": {
+    "'": 31,
+    "-": 25,
+    "0": 32,
+    "1": 29,
+    "2": 30,
+    "3": 34,
+    "4": 36,
+    "5": 33,
+    "6": 38,
+    "7": 40,
+    "8": 41,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 42,
+    "d": 26,
+    "e": 9,
+    "f": 20,
+    "g": 27,
+    "h": 18,
+    "i": 12,
+    "j": 22,
+    "k": 17,
+    "l": 21,
+    "m": 8,
+    "n": 11,
+    "o": 6,
+    "p": 28,
+    "r": 10,
+    "s": 14,
+    "t": 7,
+    "u": 16,
+    "v": 39,
+    "w": 19,
+    "y": 13,
+    "z": 35,
+    "|": 4,
+    "õ": 23,
+    "ʼ": 24
+  },
+  "azb": {
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "آ": 24,
+    "أ": 41,
+    "ؤ": 23,
+    "ئ": 9,
+    "ا": 5,
+    "ب": 16,
+    "ت": 17,
+    "ث": 37,
+    "ج": 27,
+    "ح": 29,
+    "خ": 26,
+    "د": 12,
+    "ذ": 39,
+    "ر": 10,
+    "ز": 19,
+    "س": 15,
+    "ش": 21,
+    "ص": 34,
+    "ض": 35,
+    "ط": 33,
+    "ظ": 36,
+    "ع": 28,
+    "غ": 30,
+    "ف": 32,
+    "ق": 20,
+    "ل": 11,
+    "م": 14,
+    "ن": 8,
+    "ه": 13,
+    "و": 7,
+    "پ": 31,
+    "چ": 25,
+    "ژ": 40,
+    "ک": 18,
+    "گ": 22,
+    "ی": 6
+  },
+  "azg": {
+    "'": 7,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 34,
+    "c": 16,
+    "d": 18,
+    "e": 13,
+    "f": 38,
+    "g": 26,
+    "h": 15,
+    "i": 9,
+    "j": 11,
+    "k": 41,
+    "l": 29,
+    "m": 19,
+    "n": 4,
+    "o": 8,
+    "p": 33,
+    "q": 20,
+    "r": 27,
+    "s": 17,
+    "t": 10,
+    "u": 12,
+    "v": 23,
+    "x": 25,
+    "y": 14,
+    "z": 40,
+    "|": 5,
+    "á": 31,
+    "é": 37,
+    "ë": 22,
+    "í": 32,
+    "ñ": 21,
+    "ó": 39,
+    "ö": 28,
+    "ú": 30,
+    "ü": 24,
+    "́": 35,
+    "—": 36
+  },
+  "azj-script_cyrillic": {
+    "-": 34,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "4": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 17,
+    "в": 27,
+    "г": 24,
+    "д": 11,
+    "е": 20,
+    "ж": 6,
+    "з": 16,
+    "и": 7,
+    "к": 23,
+    "л": 10,
+    "м": 13,
+    "н": 8,
+    "о": 22,
+    "п": 32,
+    "р": 9,
+    "с": 14,
+    "т": 19,
+    "у": 18,
+    "ф": 33,
+    "х": 30,
+    "ц": 28,
+    "ь": 21,
+    "э": 12,
+    "ю": 26,
+    "ј": 15,
+    "ҝ": 29,
+    "ҹ": 31,
+    "һ": 25,
+    "–": 35
+  },
+  "azj-script_latin": {
+    "!": 69,
+    "\"": 53,
+    "$": 59,
+    "%": 55,
+    "'": 56,
+    "+": 58,
+    ",": 52,
+    "-": 36,
+    ".": 49,
+    "/": 54,
+    "0": 37,
+    "1": 38,
+    "2": 40,
+    "3": 45,
+    "4": 42,
+    "5": 43,
+    "6": 47,
+    "7": 48,
+    "8": 46,
+    "9": 41,
+    ":": 50,
+    ";": 68,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 61,
+    "[": 62,
+    "]": 63,
+    "a": 5,
+    "b": 19,
+    "c": 29,
+    "d": 11,
+    "e": 17,
+    "f": 31,
+    "g": 32,
+    "h": 27,
+    "i": 7,
+    "j": 44,
+    "k": 18,
+    "l": 10,
+    "m": 15,
+    "n": 8,
+    "o": 21,
+    "p": 34,
+    "q": 23,
+    "r": 9,
+    "s": 13,
+    "t": 12,
+    "u": 20,
+    "v": 24,
+    "w": 51,
+    "x": 30,
+    "y": 16,
+    "z": 26,
+    "|": 4,
+    "°": 57,
+    "²": 67,
+    "ç": 28,
+    "õ": 60,
+    "ö": 33,
+    "ú": 66,
+    "ü": 22,
+    "ğ": 35,
+    "ı": 14,
+    "ş": 25,
+    "ə": 6,
+    "̇": 39,
+    "–": 65,
+    "—": 64
+  },
+  "azz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 11,
+    "d": 21,
+    "e": 9,
+    "f": 32,
+    "g": 28,
+    "h": 13,
+    "i": 6,
+    "j": 14,
+    "k": 35,
+    "l": 17,
+    "m": 15,
+    "n": 7,
+    "o": 12,
+    "p": 20,
+    "q": 18,
+    "r": 24,
+    "s": 16,
+    "t": 10,
+    "u": 8,
+    "v": 33,
+    "x": 25,
+    "y": 19,
+    "z": 22,
+    "|": 4,
+    "á": 30,
+    "é": 31,
+    "í": 23,
+    "ñ": 34,
+    "ó": 26,
+    "ú": 29
+  },
+  "bak": {
+    "-": 37,
+    "1": 52,
+    "8": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 54,
+    "i": 49,
+    "j": 50,
+    "|": 4,
+    "а": 5,
+    "б": 15,
+    "в": 40,
+    "г": 31,
+    "д": 24,
+    "е": 9,
+    "ж": 42,
+    "з": 39,
+    "и": 13,
+    "й": 19,
+    "к": 16,
+    "л": 10,
+    "м": 14,
+    "н": 7,
+    "о": 26,
+    "п": 28,
+    "р": 11,
+    "с": 21,
+    "т": 12,
+    "у": 17,
+    "ф": 38,
+    "х": 33,
+    "ц": 44,
+    "ч": 45,
+    "ш": 22,
+    "щ": 47,
+    "ъ": 46,
+    "ы": 8,
+    "ь": 43,
+    "э": 35,
+    "ю": 41,
+    "я": 32,
+    "ё": 48,
+    "ғ": 27,
+    "ҙ": 20,
+    "ҡ": 18,
+    "ң": 30,
+    "ҫ": 34,
+    "ү": 29,
+    "һ": 23,
+    "ә": 6,
+    "ө": 25,
+    "–": 51,
+    "—": 36
+  },
+  "bam": {
+    "'": 26,
+    "-": 32,
+    "1": 37,
+    "8": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 28,
+    "d": 22,
+    "e": 10,
+    "f": 24,
+    "g": 23,
+    "h": 29,
+    "i": 7,
+    "j": 25,
+    "k": 8,
+    "l": 12,
+    "m": 16,
+    "n": 6,
+    "o": 14,
+    "p": 30,
+    "q": 36,
+    "r": 21,
+    "s": 18,
+    "t": 20,
+    "u": 13,
+    "v": 33,
+    "w": 19,
+    "y": 15,
+    "z": 31,
+    "|": 4,
+    "ŋ": 34,
+    "ɔ": 11,
+    "ɛ": 9,
+    "ɲ": 27,
+    "–": 35
+  },
+  "ban": {
+    "'": 27,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 25,
+    "d": 16,
+    "e": 8,
+    "g": 9,
+    "h": 18,
+    "i": 7,
+    "j": 24,
+    "k": 12,
+    "l": 19,
+    "m": 17,
+    "n": 6,
+    "o": 21,
+    "p": 15,
+    "r": 14,
+    "s": 10,
+    "t": 13,
+    "u": 11,
+    "w": 22,
+    "y": 20,
+    "|": 5
+  },
+  "bao": {
+    "'": 37,
+    "1": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 9,
+    "d": 25,
+    "e": 8,
+    "f": 39,
+    "g": 26,
+    "h": 40,
+    "i": 6,
+    "j": 12,
+    "k": 46,
+    "l": 32,
+    "m": 14,
+    "n": 17,
+    "o": 11,
+    "p": 16,
+    "q": 28,
+    "r": 7,
+    "s": 27,
+    "t": 13,
+    "u": 15,
+    "v": 42,
+    "w": 21,
+    "x": 45,
+    "y": 24,
+    "z": 43,
+    "|": 4,
+    "á": 36,
+    "ã": 18,
+    "é": 30,
+    "í": 33,
+    "ñ": 23,
+    "ó": 41,
+    "õ": 29,
+    "ú": 35,
+    "ĩ": 22,
+    "ũ": 34,
+    "ʉ": 10,
+    "́": 47,
+    "̃": 20,
+    "ẽ": 38,
+    "—": 31
+  },
+  "bas": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "d": 21,
+    "e": 9,
+    "f": 28,
+    "g": 18,
+    "h": 20,
+    "i": 6,
+    "j": 26,
+    "k": 16,
+    "l": 11,
+    "m": 8,
+    "n": 7,
+    "o": 15,
+    "p": 25,
+    "r": 27,
+    "s": 23,
+    "t": 22,
+    "u": 14,
+    "w": 24,
+    "y": 19,
+    "|": 4,
+    "é": 13,
+    "ñ": 12,
+    "ô": 17
+  },
+  "bav": {
+    "'": 22,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "d": 28,
+    "e": 15,
+    "f": 21,
+    "g": 26,
+    "h": 24,
+    "i": 8,
+    "j": 33,
+    "k": 16,
+    "l": 23,
+    "m": 18,
+    "n": 9,
+    "o": 29,
+    "p": 39,
+    "r": 36,
+    "s": 12,
+    "t": 17,
+    "u": 13,
+    "v": 14,
+    "w": 11,
+    "y": 19,
+    "z": 32,
+    "|": 4,
+    "à": 35,
+    "è": 38,
+    "ì": 34,
+    "ò": 40,
+    "ù": 30,
+    "ú": 41,
+    "ŋ": 7,
+    "ɔ": 20,
+    "ə": 5,
+    "ɨ": 10,
+    "̀": 31,
+    "́": 27,
+    "̰": 42
+  },
+  "bba": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "d": 21,
+    "e": 15,
+    "f": 31,
+    "g": 20,
+    "h": 30,
+    "i": 8,
+    "k": 11,
+    "l": 26,
+    "m": 13,
+    "n": 7,
+    "o": 17,
+    "p": 24,
+    "r": 14,
+    "s": 10,
+    "t": 22,
+    "u": 6,
+    "w": 18,
+    "y": 19,
+    "|": 4,
+    "à": 33,
+    "ã": 25,
+    "è": 34,
+    "ì": 35,
+    "ò": 37,
+    "ù": 29,
+    "ĩ": 28,
+    "ũ": 36,
+    "ǹ": 27,
+    "ɔ": 9,
+    "ɛ": 16,
+    "̀": 32,
+    "̃": 23
+  },
+  "bbb": {
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "d": 19,
+    "e": 5,
+    "f": 14,
+    "g": 18,
+    "i": 7,
+    "j": 12,
+    "k": 13,
+    "l": 24,
+    "m": 15,
+    "n": 10,
+    "o": 9,
+    "r": 11,
+    "s": 20,
+    "t": 21,
+    "u": 8,
+    "v": 16,
+    "z": 22,
+    "|": 4
+  },
+  "bbc": {
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 29,
+    "d": 10,
+    "e": 19,
+    "f": 25,
+    "g": 11,
+    "h": 14,
+    "i": 7,
+    "j": 22,
+    "k": 21,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 8,
+    "p": 18,
+    "r": 16,
+    "s": 12,
+    "t": 15,
+    "u": 9,
+    "v": 23,
+    "w": 26,
+    "y": 24,
+    "z": 27,
+    "|": 4,
+    "é": 28,
+    "—": 31
+  },
+  "bbo": {
+    "'": 27,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 16,
+    "e": 19,
+    "f": 30,
+    "g": 26,
+    "h": 34,
+    "i": 9,
+    "j": 31,
+    "k": 11,
+    "l": 29,
+    "m": 10,
+    "n": 7,
+    "o": 22,
+    "p": 25,
+    "r": 12,
+    "s": 18,
+    "t": 14,
+    "u": 20,
+    "w": 28,
+    "y": 24,
+    "|": 4,
+    "á": 32,
+    "ŋ": 13,
+    "ɔ": 17,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ɲ": 23,
+    "ʋ": 15,
+    "́": 33,
+    "̣": 37,
+    "—": 36
+  },
+  "bcc-script_arabic": {
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "آ": 26,
+    "ئ": 18,
+    "ا": 5,
+    "ب": 16,
+    "ت": 7,
+    "ج": 28,
+    "د": 14,
+    "ر": 9,
+    "ز": 27,
+    "س": 17,
+    "ش": 20,
+    "ل": 23,
+    "م": 12,
+    "ن": 6,
+    "ه": 10,
+    "و": 11,
+    "ٹ": 31,
+    "پ": 22,
+    "چ": 24,
+    "ڈ": 29,
+    "ڑ": 30,
+    "ژ": 32,
+    "ک": 13,
+    "گ": 15,
+    "ۆ": 25,
+    "ی": 8,
+    "ێ": 21,
+    "ے": 19
+  },
+  "bcc-script_latin": {
+    "'": 30,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 25,
+    "d": 17,
+    "e": 16,
+    "g": 18,
+    "h": 8,
+    "i": 14,
+    "j": 28,
+    "k": 13,
+    "l": 24,
+    "m": 15,
+    "n": 7,
+    "o": 12,
+    "p": 23,
+    "r": 10,
+    "s": 11,
+    "t": 9,
+    "u": 29,
+    "w": 22,
+    "y": 20,
+    "z": 27,
+    "|": 4,
+    "á": 6,
+    "é": 19,
+    "ó": 26
+  },
+  "bcl": {
+    "'": 31,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 13,
+    "e": 22,
+    "f": 27,
+    "g": 8,
+    "h": 21,
+    "i": 7,
+    "j": 24,
+    "k": 11,
+    "l": 16,
+    "m": 12,
+    "n": 6,
+    "o": 9,
+    "p": 17,
+    "q": 32,
+    "r": 19,
+    "s": 10,
+    "t": 14,
+    "u": 15,
+    "v": 30,
+    "w": 23,
+    "x": 33,
+    "y": 20,
+    "z": 29,
+    "|": 5,
+    "—": 28
+  },
+  "bcw": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 27,
+    "d": 20,
+    "e": 23,
+    "f": 29,
+    "g": 16,
+    "h": 8,
+    "i": 7,
+    "j": 31,
+    "k": 13,
+    "l": 18,
+    "m": 10,
+    "n": 9,
+    "p": 26,
+    "r": 22,
+    "s": 11,
+    "t": 14,
+    "v": 21,
+    "w": 12,
+    "y": 15,
+    "z": 17,
+    "|": 4,
+    "ŋ": 25,
+    "ɓ": 28,
+    "ɗ": 30,
+    "ə": 6
+  },
+  "bdg": {
+    "'": 16,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 9,
+    "e": 15,
+    "g": 12,
+    "h": 23,
+    "i": 6,
+    "j": 25,
+    "k": 17,
+    "l": 13,
+    "m": 10,
+    "n": 7,
+    "o": 21,
+    "p": 19,
+    "r": 20,
+    "s": 11,
+    "t": 14,
+    "u": 8,
+    "w": 26,
+    "y": 22,
+    "|": 4
+  },
+  "bdh": {
+    "'": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 11,
+    "b": 17,
+    "c": 34,
+    "d": 10,
+    "e": 7,
+    "f": 36,
+    "g": 13,
+    "h": 38,
+    "i": 28,
+    "k": 15,
+    "l": 24,
+    "m": 18,
+    "n": 6,
+    "o": 9,
+    "p": 33,
+    "r": 20,
+    "s": 23,
+    "t": 31,
+    "u": 30,
+    "v": 39,
+    "w": 32,
+    "y": 16,
+    "z": 25,
+    "|": 4,
+    "á": 19,
+    "é": 12,
+    "í": 21,
+    "ó": 27,
+    "ú": 29,
+    "ɨ": 5,
+    "́": 8,
+    "ṇ": 35,
+    "ṛ": 37,
+    "ṿ": 40,
+    "ị": 14,
+    "ụ": 26,
+    "ꞌ": 22
+  },
+  "bdq": {
+    "'": 26,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 17,
+    "c": 34,
+    "d": 25,
+    "e": 18,
+    "g": 11,
+    "h": 7,
+    "i": 8,
+    "j": 33,
+    "k": 9,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 14,
+    "p": 21,
+    "r": 22,
+    "s": 20,
+    "t": 12,
+    "u": 19,
+    "w": 36,
+    "y": 30,
+    "|": 4,
+    "â": 37,
+    "ê": 32,
+    "ô": 27,
+    "ă": 16,
+    "đ": 28,
+    "ĕ": 29,
+    "ĭ": 39,
+    "ŏ": 35,
+    "ŭ": 38,
+    "ơ": 5,
+    "ư": 24,
+    "̆": 23
+  },
+  "bdu": {
+    "'": 23,
+    "-": 26,
+    "0": 33,
+    "1": 34,
+    "2": 37,
+    "3": 45,
+    "4": 38,
+    "5": 43,
+    "6": 40,
+    "7": 35,
+    "8": 41,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 8,
+    "c": 31,
+    "d": 25,
+    "e": 12,
+    "f": 24,
+    "g": 20,
+    "h": 30,
+    "i": 9,
+    "j": 21,
+    "k": 14,
+    "l": 32,
+    "m": 13,
+    "n": 7,
+    "o": 6,
+    "p": 29,
+    "r": 28,
+    "s": 15,
+    "t": 17,
+    "u": 18,
+    "v": 36,
+    "w": 22,
+    "x": 47,
+    "y": 19,
+    "z": 39,
+    "|": 4,
+    "ó": 46,
+    "ŋ": 27,
+    "ɔ": 16,
+    "ɗ": 10,
+    "ɛ": 11,
+    "–": 44
+  },
+  "bdv": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଁ": 31,
+    "ଂ": 41,
+    "ଃ": 7,
+    "ଅ": 29,
+    "ଆ": 23,
+    "ଇ": 33,
+    "ଉ": 25,
+    "ଏ": 36,
+    "କ": 10,
+    "ଗ": 24,
+    "ଙ": 35,
+    "ଚ": 26,
+    "ଜ": 20,
+    "ଞ": 39,
+    "ଟ": 27,
+    "ଡ": 28,
+    "ଣ": 32,
+    "ତ": 13,
+    "ଦ": 21,
+    "ନ": 17,
+    "ପ": 34,
+    "ବ": 19,
+    "ମ": 14,
+    "ର": 8,
+    "ଲ": 18,
+    "ଳ": 30,
+    "ସ": 12,
+    "ହ": 22,
+    "ା": 6,
+    "ି": 9,
+    "ୁ": 16,
+    "େ": 11,
+    "୍": 5,
+    "ୟ": 15,
+    "ୱ": 37,
+    "‍": 38
+  },
+  "beh": {
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 10,
+    "c": 26,
+    "d": 16,
+    "e": 13,
+    "f": 27,
+    "g": 20,
+    "h": 14,
+    "i": 7,
+    "k": 24,
+    "l": 29,
+    "m": 12,
+    "n": 8,
+    "o": 19,
+    "p": 22,
+    "r": 21,
+    "s": 15,
+    "t": 17,
+    "u": 9,
+    "w": 18,
+    "y": 11,
+    "|": 4,
+    "á": 30,
+    "é": 32,
+    "í": 28,
+    "ó": 33,
+    "ú": 31,
+    "ǝ": 5,
+    "́": 23
+  },
+  "bel": {
+    "!": 77,
+    "%": 80,
+    "'": 41,
+    "+": 84,
+    ",": 63,
+    "-": 40,
+    ".": 44,
+    "/": 69,
+    "0": 39,
+    "1": 42,
+    "2": 43,
+    "3": 47,
+    "4": 52,
+    "5": 46,
+    "6": 49,
+    "7": 51,
+    "8": 50,
+    "9": 45,
+    ":": 53,
+    ";": 78,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 91,
+    "[": 82,
+    "]": 83,
+    "a": 54,
+    "b": 71,
+    "c": 59,
+    "d": 68,
+    "e": 57,
+    "f": 87,
+    "g": 67,
+    "h": 70,
+    "i": 55,
+    "j": 92,
+    "k": 74,
+    "l": 60,
+    "m": 72,
+    "n": 61,
+    "o": 56,
+    "p": 64,
+    "r": 58,
+    "s": 62,
+    "t": 65,
+    "u": 86,
+    "v": 66,
+    "x": 85,
+    "y": 73,
+    "z": 88,
+    "|": 4,
+    "«": 38,
+    "°": 79,
+    "²": 75,
+    "»": 37,
+    "õ": 89,
+    "ú": 90,
+    "а": 5,
+    "б": 26,
+    "в": 21,
+    "г": 25,
+    "д": 17,
+    "е": 12,
+    "ж": 33,
+    "з": 23,
+    "и": 76,
+    "й": 31,
+    "к": 13,
+    "л": 14,
+    "м": 18,
+    "н": 6,
+    "о": 15,
+    "п": 19,
+    "р": 9,
+    "с": 10,
+    "т": 11,
+    "у": 22,
+    "ф": 36,
+    "х": 30,
+    "ц": 20,
+    "ч": 29,
+    "ш": 32,
+    "ы": 8,
+    "ь": 27,
+    "э": 28,
+    "ю": 34,
+    "я": 16,
+    "ё": 35,
+    "і": 7,
+    "ў": 24,
+    "ґ": 94,
+    "‒": 93,
+    "–": 81,
+    "—": 48
+  },
+  "bem": {
+    "'": 26,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 22,
+    "d": 24,
+    "e": 8,
+    "f": 21,
+    "g": 23,
+    "h": 20,
+    "i": 6,
+    "j": 33,
+    "k": 11,
+    "l": 9,
+    "m": 12,
+    "n": 10,
+    "o": 14,
+    "p": 19,
+    "r": 25,
+    "s": 16,
+    "t": 17,
+    "u": 7,
+    "w": 15,
+    "y": 18,
+    "|": 4,
+    "ā": 28,
+    "ē": 29,
+    "ī": 30,
+    "ŋ": 32,
+    "ō": 31,
+    "ū": 27,
+    "–": 35
+  },
+  "ben": {
+    "!": 115,
+    "\"": 46,
+    "%": 113,
+    "'": 71,
+    "+": 124,
+    ",": 70,
+    "-": 55,
+    ".": 75,
+    "/": 85,
+    "0": 48,
+    "1": 52,
+    "2": 57,
+    "3": 65,
+    "4": 63,
+    "5": 62,
+    "6": 68,
+    "7": 72,
+    "8": 69,
+    "9": 61,
+    ":": 79,
+    ";": 108,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 131,
+    "[": 127,
+    "]": 128,
+    "a": 73,
+    "b": 105,
+    "c": 77,
+    "d": 99,
+    "e": 101,
+    "f": 114,
+    "g": 104,
+    "h": 100,
+    "i": 89,
+    "j": 132,
+    "k": 110,
+    "l": 103,
+    "m": 86,
+    "n": 92,
+    "o": 97,
+    "p": 87,
+    "q": 122,
+    "r": 98,
+    "s": 84,
+    "t": 83,
+    "u": 96,
+    "v": 102,
+    "w": 106,
+    "x": 111,
+    "y": 109,
+    "z": 126,
+    "|": 4,
+    "£": 129,
+    "¥": 117,
+    "°": 121,
+    "á": 134,
+    "í": 123,
+    "õ": 118,
+    "ú": 133,
+    "ü": 125,
+    "।": 50,
+    "ঁ": 54,
+    "ং": 36,
+    "ঃ": 76,
+    "অ": 34,
+    "আ": 35,
+    "ই": 31,
+    "ঈ": 107,
+    "উ": 44,
+    "ঊ": 119,
+    "ঋ": 120,
+    "এ": 23,
+    "ঐ": 82,
+    "ও": 41,
+    "ঔ": 116,
+    "ক": 11,
+    "খ": 43,
+    "গ": 27,
+    "ঘ": 51,
+    "ঙ": 59,
+    "চ": 38,
+    "ছ": 29,
+    "জ": 26,
+    "ঝ": 66,
+    "ঞ": 56,
+    "ট": 20,
+    "ঠ": 53,
+    "ড": 39,
+    "ঢ": 94,
+    "ণ": 42,
+    "ত": 14,
+    "থ": 32,
+    "দ": 22,
+    "ধ": 40,
+    "ন": 10,
+    "প": 17,
+    "ফ": 45,
+    "ব": 13,
+    "ভ": 33,
+    "ম": 19,
+    "য": 12,
+    "র": 6,
+    "ল": 16,
+    "শ": 28,
+    "ষ": 37,
+    "স": 15,
+    "হ": 24,
+    "়": 18,
+    "া": 5,
+    "ি": 9,
+    "ী": 30,
+    "ু": 21,
+    "ূ": 47,
+    "ৃ": 49,
+    "ে": 7,
+    "ৈ": 58,
+    "ো": 25,
+    "ৌ": 60,
+    "্": 8,
+    "ৎ": 64,
+    "০": 67,
+    "১": 74,
+    "২": 80,
+    "৩": 93,
+    "৪": 78,
+    "৫": 90,
+    "৬": 88,
+    "৭": 95,
+    "৮": 91,
+    "৯": 81,
+    "৷": 130,
+    "‍": 112,
+    "—": 135
+  },
+  "bep": {
+    "'": 28,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 29,
+    "d": 22,
+    "e": 11,
+    "g": 20,
+    "h": 16,
+    "i": 6,
+    "j": 27,
+    "k": 17,
+    "l": 18,
+    "m": 14,
+    "n": 8,
+    "o": 10,
+    "p": 15,
+    "r": 19,
+    "s": 24,
+    "t": 12,
+    "u": 9,
+    "w": 23,
+    "|": 4,
+    "á": 13,
+    "ó": 25,
+    "ꞌ": 7
+  },
+  "bex": {
+    "'": 17,
+    "2": 35,
+    "3": 33,
+    "8": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 31,
+    "d": 11,
+    "e": 9,
+    "g": 20,
+    "h": 32,
+    "i": 6,
+    "j": 29,
+    "k": 7,
+    "l": 18,
+    "m": 12,
+    "n": 10,
+    "o": 14,
+    "p": 27,
+    "r": 22,
+    "s": 28,
+    "t": 16,
+    "u": 24,
+    "w": 30,
+    "y": 19,
+    "z": 25,
+    "|": 4,
+    "ë": 23,
+    "ï": 15,
+    "ö": 21,
+    "ŋ": 26,
+    "ɔ": 8
+  },
+  "bfa": {
+    "'": 25,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "d": 16,
+    "e": 10,
+    "g": 17,
+    "h": 30,
+    "i": 7,
+    "j": 24,
+    "k": 9,
+    "l": 13,
+    "m": 20,
+    "n": 6,
+    "o": 8,
+    "p": 23,
+    "r": 19,
+    "s": 22,
+    "t": 12,
+    "u": 11,
+    "w": 21,
+    "y": 14,
+    "|": 4,
+    "é": 29,
+    "ö": 18,
+    "ŋ": 15,
+    "—": 28
+  },
+  "bfo": {
+    "-": 20,
+    "1": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 32,
+    "d": 29,
+    "e": 15,
+    "f": 30,
+    "g": 35,
+    "h": 44,
+    "i": 21,
+    "j": 26,
+    "k": 24,
+    "l": 14,
+    "m": 16,
+    "n": 7,
+    "o": 27,
+    "p": 22,
+    "r": 11,
+    "s": 19,
+    "t": 23,
+    "u": 25,
+    "v": 38,
+    "w": 17,
+    "y": 18,
+    "|": 4,
+    "á": 41,
+    "ã": 28,
+    "é": 49,
+    "õ": 37,
+    "ú": 46,
+    "ĩ": 43,
+    "ı": 51,
+    "ŋ": 33,
+    "ũ": 39,
+    "ƴ": 31,
+    "ɓ": 45,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ʋ": 9,
+    "ʻ": 47,
+    "ʼ": 34,
+    "́": 42,
+    "̃": 13,
+    "ẽ": 36,
+    "—": 48,
+    "ↄ": 12,
+    "ꞌ": 40
+  },
+  "bfy": {
+    "'": 54,
+    "-": 47,
+    "1": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 24,
+    "ं": 16,
+    "अ": 23,
+    "आ": 29,
+    "इ": 22,
+    "ई": 35,
+    "उ": 20,
+    "ऊ": 37,
+    "ए": 36,
+    "ओ": 49,
+    "क": 8,
+    "ख": 32,
+    "ग": 33,
+    "घ": 44,
+    "च": 30,
+    "छ": 45,
+    "ज": 25,
+    "झ": 52,
+    "ट": 48,
+    "ठ": 43,
+    "ड": 39,
+    "ढ": 51,
+    "त": 15,
+    "थ": 46,
+    "द": 26,
+    "ध": 42,
+    "न": 10,
+    "प": 18,
+    "फ": 50,
+    "ब": 17,
+    "भ": 38,
+    "म": 12,
+    "य": 19,
+    "र": 9,
+    "ल": 21,
+    "व": 40,
+    "स": 13,
+    "ह": 7,
+    "़": 41,
+    "ा": 5,
+    "ि": 11,
+    "ी": 14,
+    "ु": 28,
+    "ू": 34,
+    "े": 6,
+    "ै": 53,
+    "ो": 31,
+    "ौ": 55,
+    "्": 27
+  },
+  "bfz": {
+    "-": 54,
+    "4": 65,
+    "8": 64,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 63,
+    "ं": 25,
+    "ः": 61,
+    "अ": 53,
+    "आ": 19,
+    "इ": 38,
+    "ई": 32,
+    "उ": 44,
+    "ऊ": 36,
+    "ए": 46,
+    "ऐ": 37,
+    "ओ": 55,
+    "औ": 29,
+    "क": 12,
+    "ख": 43,
+    "ग": 35,
+    "घ": 57,
+    "च": 41,
+    "छ": 45,
+    "ज": 24,
+    "झ": 49,
+    "ञ": 60,
+    "ट": 48,
+    "ठ": 56,
+    "ड": 40,
+    "ढ": 59,
+    "ण": 27,
+    "त": 16,
+    "थ": 39,
+    "द": 18,
+    "ध": 51,
+    "न": 26,
+    "प": 23,
+    "फ": 50,
+    "ब": 20,
+    "भ": 42,
+    "म": 15,
+    "य": 30,
+    "र": 6,
+    "ल": 14,
+    "व": 31,
+    "श": 28,
+    "ष": 58,
+    "स": 17,
+    "ह": 34,
+    "़": 11,
+    "ा": 5,
+    "ि": 13,
+    "ी": 9,
+    "ु": 10,
+    "ू": 52,
+    "ृ": 62,
+    "े": 22,
+    "ै": 7,
+    "ो": 33,
+    "ौ": 8,
+    "्": 21,
+    "‍": 47
+  },
+  "bgc": {
+    "'": 61,
+    "-": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 38,
+    "ं": 25,
+    "अ": 28,
+    "आ": 40,
+    "इ": 44,
+    "ई": 49,
+    "उ": 31,
+    "ऊ": 57,
+    "ऋ": 65,
+    "ए": 41,
+    "ऐ": 64,
+    "ओ": 53,
+    "औ": 59,
+    "क": 7,
+    "ख": 34,
+    "ग": 30,
+    "घ": 52,
+    "च": 35,
+    "छ": 48,
+    "ज": 21,
+    "झ": 55,
+    "ञ": 60,
+    "ट": 45,
+    "ठ": 51,
+    "ड": 39,
+    "ढ": 56,
+    "ण": 26,
+    "त": 13,
+    "थ": 33,
+    "द": 24,
+    "ध": 43,
+    "न": 14,
+    "प": 20,
+    "फ": 47,
+    "ब": 23,
+    "भ": 37,
+    "म": 12,
+    "य": 17,
+    "र": 6,
+    "ल": 27,
+    "ळ": 46,
+    "व": 22,
+    "श": 36,
+    "ष": 58,
+    "स": 9,
+    "ह": 15,
+    "़": 42,
+    "ा": 5,
+    "ि": 19,
+    "ी": 16,
+    "ु": 29,
+    "ू": 32,
+    "ृ": 63,
+    "े": 10,
+    "ै": 8,
+    "ो": 18,
+    "ौ": 54,
+    "्": 11,
+    "‍": 62
+  },
+  "bgq": {
+    "'": 49,
+    "-": 57,
+    "0": 58,
+    "1": 61,
+    "2": 60,
+    "3": 63,
+    "4": 59,
+    "5": 62,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 12,
+    "अ": 29,
+    "आ": 28,
+    "इ": 45,
+    "ई": 34,
+    "उ": 52,
+    "ऊ": 38,
+    "ए": 46,
+    "ऐ": 56,
+    "ओ": 47,
+    "क": 11,
+    "ख": 33,
+    "ग": 10,
+    "घ": 53,
+    "च": 35,
+    "छ": 50,
+    "ज": 27,
+    "झ": 40,
+    "ट": 43,
+    "ठ": 51,
+    "ड": 36,
+    "ढ": 54,
+    "ण": 26,
+    "त": 20,
+    "थ": 37,
+    "द": 23,
+    "ध": 44,
+    "न": 17,
+    "प": 21,
+    "फ": 48,
+    "ब": 16,
+    "भ": 39,
+    "म": 14,
+    "य": 18,
+    "र": 6,
+    "ल": 25,
+    "ळ": 41,
+    "व": 31,
+    "स": 13,
+    "ह": 22,
+    "़": 42,
+    "ा": 5,
+    "ि": 19,
+    "ी": 15,
+    "ु": 32,
+    "ू": 30,
+    "े": 9,
+    "ै": 7,
+    "ो": 8,
+    "ौ": 55,
+    "्": 24
+  },
+  "bgr": {
+    "'": 33,
+    "-": 34,
+    "1": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 19,
+    "d": 22,
+    "e": 13,
+    "f": 27,
+    "g": 15,
+    "h": 6,
+    "i": 8,
+    "j": 31,
+    "k": 12,
+    "l": 10,
+    "m": 14,
+    "n": 7,
+    "o": 23,
+    "p": 21,
+    "q": 38,
+    "r": 20,
+    "s": 16,
+    "t": 11,
+    "u": 9,
+    "v": 28,
+    "w": 17,
+    "y": 37,
+    "z": 24,
+    "|": 4,
+    "à": 35,
+    "â": 18,
+    "ê": 25,
+    "î": 32,
+    "û": 29,
+    "ṭ": 30
+  },
+  "bgt": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 26,
+    "d": 18,
+    "e": 8,
+    "f": 24,
+    "g": 10,
+    "h": 12,
+    "i": 6,
+    "j": 23,
+    "k": 11,
+    "l": 19,
+    "m": 15,
+    "n": 7,
+    "o": 9,
+    "p": 22,
+    "r": 16,
+    "s": 20,
+    "t": 14,
+    "u": 13,
+    "v": 17,
+    "w": 25,
+    "|": 4
+  },
+  "bgw": {
+    "'": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 20,
+    "अ": 31,
+    "आ": 18,
+    "इ": 36,
+    "ई": 33,
+    "उ": 35,
+    "ऊ": 43,
+    "ए": 32,
+    "ओ": 37,
+    "क": 8,
+    "ख": 38,
+    "ग": 22,
+    "च": 27,
+    "ज": 25,
+    "ट": 26,
+    "ड": 30,
+    "त": 15,
+    "द": 23,
+    "न": 12,
+    "प": 21,
+    "ब": 9,
+    "भ": 40,
+    "म": 13,
+    "य": 17,
+    "र": 6,
+    "ल": 11,
+    "व": 28,
+    "स": 16,
+    "ह": 29,
+    "़": 34,
+    "ा": 5,
+    "ि": 24,
+    "ी": 10,
+    "ु": 19,
+    "ू": 39,
+    "े": 7,
+    "ो": 14,
+    "‍": 42
+  },
+  "bha": {
+    "'": 48,
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 18,
+    "अ": 24,
+    "आ": 32,
+    "इ": 47,
+    "ई": 31,
+    "ऊ": 41,
+    "ए": 37,
+    "क": 7,
+    "ख": 33,
+    "ग": 25,
+    "घ": 40,
+    "च": 30,
+    "छ": 28,
+    "ज": 26,
+    "झ": 43,
+    "ट": 35,
+    "ठ": 39,
+    "ड": 29,
+    "ढ": 45,
+    "त": 17,
+    "थ": 46,
+    "द": 19,
+    "ध": 42,
+    "न": 11,
+    "प": 22,
+    "फ": 38,
+    "ब": 20,
+    "भ": 36,
+    "म": 16,
+    "य": 13,
+    "र": 6,
+    "ल": 21,
+    "व": 23,
+    "स": 15,
+    "ह": 10,
+    "़": 34,
+    "ा": 5,
+    "ी": 9,
+    "ु": 49,
+    "ू": 14,
+    "े": 8,
+    "ो": 12,
+    "्": 27
+  },
+  "bht": {
+    "'": 65,
+    "-": 54,
+    "0": 64,
+    "2": 69,
+    "3": 67,
+    "4": 66,
+    "5": 68,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 57,
+    "ं": 13,
+    "अ": 35,
+    "आ": 43,
+    "इ": 39,
+    "ई": 31,
+    "उ": 32,
+    "ऊ": 58,
+    "ए": 55,
+    "ऐ": 42,
+    "ओ": 50,
+    "औ": 61,
+    "क": 8,
+    "ख": 40,
+    "ग": 30,
+    "घ": 51,
+    "च": 28,
+    "छ": 45,
+    "ज": 19,
+    "झ": 47,
+    "ञ": 59,
+    "ट": 48,
+    "ठ": 49,
+    "ड": 33,
+    "ढ": 56,
+    "ण": 26,
+    "त": 18,
+    "थ": 41,
+    "द": 10,
+    "ध": 46,
+    "न": 12,
+    "प": 23,
+    "फ": 44,
+    "ब": 25,
+    "भ": 38,
+    "म": 22,
+    "य": 16,
+    "र": 7,
+    "ल": 17,
+    "व": 37,
+    "श": 29,
+    "ष": 53,
+    "स": 11,
+    "ह": 21,
+    "़": 34,
+    "ा": 5,
+    "ि": 14,
+    "ी": 9,
+    "ु": 24,
+    "ू": 36,
+    "ृ": 60,
+    "े": 6,
+    "ै": 27,
+    "ॉ": 63,
+    "ो": 15,
+    "ौ": 52,
+    "्": 20,
+    "‍": 62
+  },
+  "bhz": {
+    "'": 28,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "d": 20,
+    "e": 11,
+    "g": 21,
+    "h": 17,
+    "i": 6,
+    "k": 16,
+    "l": 18,
+    "m": 14,
+    "n": 8,
+    "o": 9,
+    "p": 15,
+    "r": 19,
+    "s": 24,
+    "t": 12,
+    "u": 10,
+    "w": 23,
+    "|": 4,
+    "á": 13,
+    "ó": 25,
+    "—": 27,
+    "ꞌ": 7
+  },
+  "bib": {
+    "'": 20,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 31,
+    "d": 22,
+    "e": 24,
+    "f": 34,
+    "g": 27,
+    "h": 25,
+    "i": 19,
+    "j": 33,
+    "k": 12,
+    "l": 28,
+    "m": 9,
+    "n": 8,
+    "o": 17,
+    "p": 32,
+    "r": 13,
+    "s": 16,
+    "t": 23,
+    "u": 29,
+    "v": 35,
+    "w": 18,
+    "y": 14,
+    "z": 26,
+    "|": 4,
+    "ŋ": 21,
+    "ɔ": 6,
+    "ə": 30,
+    "ɛ": 15,
+    "ɩ": 7,
+    "ʋ": 11
+  },
+  "bim": {
+    "'": 30,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 29,
+    "d": 22,
+    "e": 9,
+    "f": 26,
+    "g": 25,
+    "h": 27,
+    "i": 7,
+    "j": 23,
+    "k": 8,
+    "l": 21,
+    "m": 14,
+    "n": 6,
+    "o": 13,
+    "p": 20,
+    "q": 32,
+    "r": 16,
+    "s": 17,
+    "t": 11,
+    "u": 12,
+    "w": 24,
+    "y": 18,
+    "|": 4,
+    "ê": 33,
+    "ŋ": 19,
+    "ƒ": 31,
+    "ɔ": 15
+  },
+  "bis": {
+    "'": 28,
+    "-": 27,
+    "6": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "d": 24,
+    "e": 7,
+    "f": 17,
+    "g": 12,
+    "h": 23,
+    "i": 11,
+    "j": 26,
+    "k": 18,
+    "l": 8,
+    "m": 9,
+    "n": 10,
+    "o": 6,
+    "p": 21,
+    "r": 22,
+    "s": 14,
+    "t": 13,
+    "u": 16,
+    "v": 25,
+    "w": 20,
+    "y": 19,
+    "|": 4
+  },
+  "biv": {
+    "'": 27,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 32,
+    "d": 29,
+    "e": 19,
+    "f": 30,
+    "g": 34,
+    "h": 31,
+    "i": 22,
+    "j": 28,
+    "k": 17,
+    "l": 14,
+    "m": 20,
+    "n": 6,
+    "o": 24,
+    "p": 23,
+    "r": 11,
+    "s": 16,
+    "t": 18,
+    "u": 25,
+    "v": 36,
+    "w": 21,
+    "y": 13,
+    "|": 4,
+    "ã": 26,
+    "õ": 37,
+    "ĩ": 39,
+    "ŋ": 33,
+    "ũ": 35,
+    "ɔ": 9,
+    "ɛ": 8,
+    "ɩ": 7,
+    "ʋ": 12,
+    "̃": 15,
+    "ẽ": 38
+  },
+  "bjr": {
+    "'": 37,
+    "-": 29,
+    "0": 30,
+    "1": 34,
+    "2": 32,
+    "4": 35,
+    "5": 36,
+    "6": 38,
+    "7": 40,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "d": 17,
+    "e": 13,
+    "f": 14,
+    "g": 28,
+    "h": 27,
+    "i": 6,
+    "k": 16,
+    "m": 9,
+    "n": 7,
+    "o": 10,
+    "p": 22,
+    "q": 8,
+    "r": 11,
+    "s": 15,
+    "t": 18,
+    "u": 12,
+    "w": 33,
+    "y": 21,
+    "|": 5,
+    "á": 20,
+    "é": 24,
+    "í": 19,
+    "ó": 26,
+    "ú": 25,
+    "̱": 31,
+    "ṉ": 41
+  },
+  "bjv": {
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "d": 12,
+    "e": 10,
+    "g": 17,
+    "h": 33,
+    "i": 18,
+    "j": 13,
+    "k": 9,
+    "l": 14,
+    "m": 15,
+    "n": 7,
+    "o": 16,
+    "p": 30,
+    "r": 19,
+    "s": 21,
+    "t": 11,
+    "u": 23,
+    "w": 24,
+    "y": 26,
+    "|": 4,
+    "é": 31,
+    "ɓ": 27,
+    "ɔ": 20,
+    "ə": 8,
+    "ɨ": 5,
+    "̰": 29,
+    "ḛ": 28,
+    "ḭ": 32
+  },
+  "bjw": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 35,
+    "d": 32,
+    "e": 26,
+    "f": 34,
+    "g": 24,
+    "i": 12,
+    "j": 25,
+    "k": 20,
+    "l": 10,
+    "m": 19,
+    "n": 7,
+    "o": 29,
+    "p": 23,
+    "r": 18,
+    "s": 22,
+    "t": 27,
+    "u": 28,
+    "v": 37,
+    "w": 30,
+    "y": 14,
+    "|": 4,
+    "ä": 13,
+    "ë": 33,
+    "ö": 31,
+    "ü": 17,
+    "ŋ": 36,
+    "ɔ": 9,
+    "ɛ": 21,
+    "ɩ": 8,
+    "ʋ": 16,
+    "‐": 11
+  },
+  "bjz": {
+    "'": 49,
+    "-": 23,
+    "0": 32,
+    "1": 33,
+    "2": 34,
+    "3": 37,
+    "4": 35,
+    "5": 44,
+    "6": 38,
+    "7": 40,
+    "8": 43,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 45,
+    "d": 14,
+    "e": 6,
+    "f": 21,
+    "g": 12,
+    "h": 20,
+    "i": 7,
+    "j": 22,
+    "k": 19,
+    "l": 29,
+    "m": 15,
+    "n": 10,
+    "o": 8,
+    "p": 30,
+    "q": 47,
+    "r": 9,
+    "s": 18,
+    "t": 13,
+    "u": 11,
+    "v": 17,
+    "w": 41,
+    "y": 28,
+    "|": 4,
+    "á": 24,
+    "ã": 48,
+    "é": 36,
+    "í": 27,
+    "ó": 25,
+    "ú": 26,
+    "ĩ": 31,
+    "ũ": 46,
+    "ẽ": 39
+  },
+  "bkd": {
+    "'": 34,
+    "-": 25,
+    "0": 38,
+    "1": 37,
+    "2": 39,
+    "3": 42,
+    "4": 35,
+    "5": 40,
+    "6": 41,
+    "7": 44,
+    "9": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 29,
+    "d": 10,
+    "e": 15,
+    "f": 30,
+    "g": 11,
+    "h": 9,
+    "i": 8,
+    "j": 27,
+    "k": 16,
+    "l": 17,
+    "m": 14,
+    "n": 6,
+    "o": 24,
+    "p": 19,
+    "r": 23,
+    "s": 12,
+    "t": 13,
+    "u": 7,
+    "v": 33,
+    "w": 20,
+    "x": 36,
+    "y": 18,
+    "z": 32,
+    "|": 5,
+    "à": 22,
+    "è": 31,
+    "ì": 26,
+    "ù": 28
+  },
+  "bkv": {
+    "'": 11,
+    "-": 28,
+    "0": 51,
+    "7": 55,
+    "8": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 20,
+    "d": 23,
+    "e": 8,
+    "f": 26,
+    "g": 9,
+    "h": 13,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 30,
+    "m": 15,
+    "n": 7,
+    "o": 17,
+    "p": 24,
+    "r": 18,
+    "s": 21,
+    "t": 19,
+    "u": 10,
+    "w": 22,
+    "y": 16,
+    "|": 4,
+    "à": 40,
+    "á": 32,
+    "â": 53,
+    "è": 27,
+    "é": 31,
+    "ì": 29,
+    "í": 34,
+    "î": 49,
+    "ò": 42,
+    "ó": 45,
+    "ù": 43,
+    "ú": 33,
+    "û": 50,
+    "ā": 41,
+    "ē": 35,
+    "ě": 36,
+    "ī": 37,
+    "ń": 46,
+    "ō": 47,
+    "ū": 38,
+    "ǎ": 54,
+    "ǹ": 44,
+    "̀": 39,
+    "̄": 48,
+    "ḿ": 52
+  },
+  "blh": {
+    "1": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 27,
+    "d": 12,
+    "e": 20,
+    "f": 37,
+    "g": 30,
+    "i": 21,
+    "j": 23,
+    "k": 16,
+    "l": 11,
+    "m": 26,
+    "n": 5,
+    "o": 28,
+    "p": 39,
+    "s": 17,
+    "t": 32,
+    "u": 24,
+    "v": 36,
+    "w": 7,
+    "y": 29,
+    "|": 4,
+    "à": 10,
+    "á": 34,
+    "è": 15,
+    "é": 38,
+    "ì": 13,
+    "í": 22,
+    "ò": 19,
+    "ó": 33,
+    "ù": 18,
+    "ú": 35,
+    "ě": 44,
+    "ŋ": 31,
+    "ŏ": 45,
+    "ǎ": 46,
+    "ǐ": 40,
+    "ǔ": 43,
+    "ɔ": 6,
+    "ɛ": 9,
+    "ɤ": 41,
+    "̀": 8,
+    "́": 25,
+    "̌": 42
+  },
+  "blt": {
+    "'": 5,
+    "0": 63,
+    "1": 64,
+    "2": 66,
+    "3": 68,
+    "4": 65,
+    "5": 70,
+    "6": 71,
+    "7": 69,
+    "8": 72,
+    "9": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 35,
+    "c": 8,
+    "d": 33,
+    "e": 51,
+    "g": 12,
+    "h": 7,
+    "i": 14,
+    "k": 20,
+    "l": 21,
+    "m": 13,
+    "n": 6,
+    "o": 22,
+    "p": 11,
+    "q": 47,
+    "s": 19,
+    "t": 15,
+    "u": 10,
+    "v": 28,
+    "y": 26,
+    "|": 4,
+    "á": 16,
+    "é": 37,
+    "ê": 24,
+    "í": 30,
+    "ó": 38,
+    "ô": 41,
+    "ú": 40,
+    "ă": 46,
+    "đ": 23,
+    "ơ": 54,
+    "ư": 18,
+    "ạ": 25,
+    "ả": 17,
+    "ắ": 34,
+    "ẳ": 52,
+    "ặ": 36,
+    "ẹ": 45,
+    "ẻ": 57,
+    "ế": 43,
+    "ể": 61,
+    "ệ": 50,
+    "ỉ": 58,
+    "ị": 49,
+    "ọ": 27,
+    "ỏ": 44,
+    "ố": 32,
+    "ổ": 60,
+    "ộ": 53,
+    "ớ": 29,
+    "ở": 62,
+    "ợ": 59,
+    "ụ": 48,
+    "ủ": 42,
+    "ứ": 39,
+    "ử": 56,
+    "ự": 55,
+    "‐": 31
+  },
+  "blx": {
+    "'": 32,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 26,
+    "d": 20,
+    "e": 22,
+    "f": 28,
+    "g": 15,
+    "h": 27,
+    "i": 7,
+    "j": 25,
+    "k": 12,
+    "l": 13,
+    "m": 11,
+    "n": 6,
+    "o": 23,
+    "p": 16,
+    "q": 31,
+    "r": 21,
+    "s": 10,
+    "t": 14,
+    "u": 8,
+    "v": 29,
+    "w": 19,
+    "x": 33,
+    "y": 9,
+    "z": 30,
+    "|": 5,
+    "ê": 18
+  },
+  "blz": {
+    "'": 11,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 29,
+    "d": 22,
+    "e": 14,
+    "f": 26,
+    "g": 17,
+    "h": 25,
+    "i": 7,
+    "j": 27,
+    "k": 12,
+    "l": 16,
+    "m": 10,
+    "n": 6,
+    "o": 8,
+    "p": 19,
+    "r": 18,
+    "s": 13,
+    "t": 15,
+    "u": 9,
+    "w": 23,
+    "y": 21,
+    "z": 28,
+    "|": 5,
+    "—": 30
+  },
+  "bmq": {
+    "'": 24,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 26,
+    "d": 22,
+    "e": 10,
+    "f": 27,
+    "g": 37,
+    "h": 20,
+    "i": 7,
+    "k": 31,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 9,
+    "p": 28,
+    "r": 14,
+    "s": 18,
+    "t": 21,
+    "u": 8,
+    "v": 29,
+    "w": 11,
+    "y": 16,
+    "z": 23,
+    "|": 4,
+    "à": 41,
+    "á": 30,
+    "è": 47,
+    "é": 46,
+    "ì": 42,
+    "í": 34,
+    "ò": 48,
+    "ó": 32,
+    "ù": 45,
+    "ú": 43,
+    "ā": 39,
+    "ē": 40,
+    "ī": 35,
+    "ō": 38,
+    "ū": 36,
+    "ɓ": 19,
+    "ɛ": 12,
+    "ɲ": 25,
+    "̀": 50,
+    "̄": 44,
+    "–": 49
+  },
+  "bmr": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 11,
+    "d": 21,
+    "e": 13,
+    "f": 22,
+    "g": 20,
+    "h": 15,
+    "i": 7,
+    "j": 14,
+    "k": 31,
+    "l": 24,
+    "m": 10,
+    "n": 12,
+    "o": 5,
+    "p": 28,
+    "q": 27,
+    "r": 18,
+    "s": 19,
+    "t": 16,
+    "u": 9,
+    "v": 23,
+    "x": 30,
+    "y": 25,
+    "z": 29,
+    "|": 6,
+    "ñ": 26,
+    "ɨ": 8
+  },
+  "bmu": {
+    "'": 34,
+    "-": 29,
+    "0": 31,
+    "1": 33,
+    "2": 37,
+    "3": 38,
+    "4": 32,
+    "5": 36,
+    "6": 39,
+    "7": 35,
+    "8": 41,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 28,
+    "e": 9,
+    "f": 30,
+    "g": 14,
+    "h": 23,
+    "i": 7,
+    "j": 24,
+    "k": 10,
+    "l": 16,
+    "m": 8,
+    "n": 12,
+    "o": 19,
+    "p": 21,
+    "q": 27,
+    "r": 17,
+    "s": 20,
+    "t": 15,
+    "u": 13,
+    "w": 22,
+    "y": 26,
+    "z": 25,
+    "|": 4,
+    "ö": 6,
+    "ŋ": 11
+  },
+  "bmv": {
+    "'": 41,
+    "-": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 34,
+    "d": 37,
+    "e": 17,
+    "f": 22,
+    "g": 13,
+    "h": 7,
+    "i": 8,
+    "j": 27,
+    "k": 10,
+    "l": 24,
+    "m": 11,
+    "n": 6,
+    "o": 28,
+    "p": 40,
+    "s": 21,
+    "t": 9,
+    "u": 15,
+    "w": 14,
+    "y": 12,
+    "|": 4,
+    "à": 16,
+    "â": 29,
+    "è": 26,
+    "ê": 33,
+    "ì": 25,
+    "î": 38,
+    "ò": 30,
+    "ô": 31,
+    "ù": 18,
+    "û": 35,
+    "ŋ": 32,
+    "ə": 23,
+    "ɨ": 19,
+    "̀": 39,
+    "̂": 36
+  },
+  "bng": {
+    "'": 63,
+    "-": 61,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "p": 64,
+    "|": 4,
+    "ঁ": 40,
+    "ং": 41,
+    "ঃ": 58,
+    "অ": 39,
+    "আ": 22,
+    "ই": 29,
+    "ঈ": 47,
+    "উ": 45,
+    "ঊ": 65,
+    "ঋ": 62,
+    "এ": 28,
+    "ঐ": 59,
+    "ও": 34,
+    "ক": 12,
+    "খ": 32,
+    "গ": 31,
+    "ঘ": 52,
+    "ঙ": 49,
+    "চ": 38,
+    "ছ": 25,
+    "জ": 26,
+    "ঝ": 57,
+    "ঞ": 54,
+    "ট": 42,
+    "ঠ": 51,
+    "ড": 44,
+    "ঢ": 60,
+    "ণ": 43,
+    "ত": 10,
+    "থ": 33,
+    "দ": 16,
+    "ধ": 36,
+    "ন": 9,
+    "প": 19,
+    "ফ": 50,
+    "ব": 14,
+    "ভ": 37,
+    "ম": 15,
+    "য": 13,
+    "র": 7,
+    "ল": 17,
+    "শ": 24,
+    "ষ": 35,
+    "স": 18,
+    "হ": 27,
+    "়": 20,
+    "া": 6,
+    "ি": 11,
+    "ী": 30,
+    "ু": 21,
+    "ূ": 46,
+    "ৃ": 48,
+    "ে": 5,
+    "ৈ": 56,
+    "ো": 23,
+    "ৌ": 53,
+    "্": 8,
+    "ৎ": 55
+  },
+  "bno": {
+    "'": 25,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 18,
+    "e": 22,
+    "f": 30,
+    "g": 8,
+    "h": 21,
+    "i": 7,
+    "j": 27,
+    "k": 9,
+    "l": 19,
+    "m": 15,
+    "n": 6,
+    "o": 10,
+    "p": 17,
+    "q": 32,
+    "r": 13,
+    "s": 11,
+    "t": 12,
+    "u": 14,
+    "v": 28,
+    "w": 23,
+    "x": 33,
+    "y": 16,
+    "z": 29,
+    "|": 5,
+    "à": 34,
+    "á": 31,
+    "é": 35
+  },
+  "bnp": {
+    "'": 25,
+    "-": 29,
+    "0": 23,
+    "1": 26,
+    "2": 27,
+    "3": 32,
+    "4": 30,
+    "5": 28,
+    "6": 33,
+    "7": 34,
+    "8": 35,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 22,
+    "e": 10,
+    "g": 18,
+    "h": 17,
+    "i": 6,
+    "j": 36,
+    "k": 12,
+    "l": 14,
+    "m": 13,
+    "n": 8,
+    "o": 11,
+    "p": 20,
+    "r": 9,
+    "s": 21,
+    "t": 15,
+    "u": 7,
+    "v": 16,
+    "|": 4,
+    "­": 24
+  },
+  "boa": {
+    "'": 46,
+    "-": 36,
+    "0": 37,
+    "1": 38,
+    "2": 41,
+    "3": 42,
+    "4": 40,
+    "5": 43,
+    "6": 44,
+    "7": 39,
+    "8": 45,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 21,
+    "c": 19,
+    "d": 20,
+    "e": 6,
+    "g": 35,
+    "h": 15,
+    "i": 17,
+    "j": 12,
+    "k": 31,
+    "l": 28,
+    "m": 10,
+    "n": 18,
+    "o": 27,
+    "p": 30,
+    "r": 25,
+    "s": 24,
+    "t": 11,
+    "u": 16,
+    "v": 23,
+    "w": 32,
+    "y": 13,
+    "|": 4,
+    "á": 5,
+    "é": 7,
+    "í": 9,
+    "ñ": 33,
+    "ó": 26,
+    "ú": 14,
+    "ɨ": 22,
+    "́": 29,
+    "—": 34
+  },
+  "bod": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "་": 4,
+    "ཀ": 23,
+    "ཁ": 21,
+    "ག": 9,
+    "ང": 11,
+    "ཅ": 26,
+    "ཆ": 27,
+    "ཇ": 44,
+    "ཉ": 36,
+    "ཊ": 59,
+    "ཌ": 60,
+    "ཎ": 61,
+    "ཏ": 34,
+    "ཐ": 30,
+    "ད": 7,
+    "ན": 10,
+    "པ": 20,
+    "ཕ": 28,
+    "བ": 14,
+    "མ": 12,
+    "ཙ": 42,
+    "ཚ": 29,
+    "ཛ": 46,
+    "ཝ": 58,
+    "ཞ": 25,
+    "ཟ": 38,
+    "འ": 18,
+    "ཡ": 22,
+    "ར": 13,
+    "ལ": 19,
+    "ཤ": 31,
+    "ས": 5,
+    "ཧ": 50,
+    "ཨ": 48,
+    "ཪ": 64,
+    "ཱ": 39,
+    "ི": 6,
+    "ུ": 16,
+    "ེ": 15,
+    "ཻ": 62,
+    "ོ": 8,
+    "ྐ": 35,
+    "ྒ": 37,
+    "ྔ": 49,
+    "ྕ": 56,
+    "ྗ": 47,
+    "ྙ": 52,
+    "ྟ": 32,
+    "ྡ": 41,
+    "ྣ": 33,
+    "ྤ": 43,
+    "ྥ": 63,
+    "ྦ": 54,
+    "ྨ": 45,
+    "ྩ": 51,
+    "ྫ": 55,
+    "ྭ": 57,
+    "ྱ": 17,
+    "ྲ": 24,
+    "ླ": 40,
+    "ྷ": 53
+  },
+  "boj": {
+    "'": 32,
+    "0": 28,
+    "1": 31,
+    "2": 33,
+    "3": 38,
+    "4": 35,
+    "5": 36,
+    "6": 37,
+    "7": 34,
+    "8": 39,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "d": 21,
+    "e": 7,
+    "f": 29,
+    "g": 11,
+    "h": 27,
+    "i": 6,
+    "j": 20,
+    "k": 23,
+    "l": 19,
+    "m": 12,
+    "n": 10,
+    "o": 9,
+    "p": 25,
+    "q": 8,
+    "r": 18,
+    "s": 13,
+    "t": 17,
+    "u": 14,
+    "v": 30,
+    "w": 24,
+    "y": 22,
+    "|": 4,
+    "ñ": 26,
+    "ŋ": 16
+  },
+  "bom": {
+    "'": 59,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 39,
+    "d": 20,
+    "e": 6,
+    "f": 36,
+    "g": 11,
+    "h": 22,
+    "i": 17,
+    "j": 27,
+    "k": 15,
+    "l": 24,
+    "m": 12,
+    "n": 7,
+    "o": 10,
+    "p": 25,
+    "r": 18,
+    "s": 16,
+    "t": 13,
+    "u": 23,
+    "v": 26,
+    "w": 14,
+    "y": 8,
+    "z": 40,
+    "|": 4,
+    "à": 29,
+    "á": 34,
+    "â": 31,
+    "è": 35,
+    "é": 28,
+    "ê": 32,
+    "ì": 58,
+    "í": 53,
+    "î": 48,
+    "ò": 60,
+    "ó": 45,
+    "ô": 41,
+    "ù": 55,
+    "ú": 56,
+    "û": 49,
+    "ā": 42,
+    "ē": 43,
+    "ě": 50,
+    "ī": 54,
+    "ō": 46,
+    "ū": 62,
+    "ǎ": 51,
+    "ǐ": 57,
+    "ǒ": 63,
+    "ǔ": 61,
+    "ɔ": 21,
+    "ɛ": 9,
+    "̀": 30,
+    "́": 33,
+    "̂": 38,
+    "̄": 47,
+    "̌": 44,
+    "—": 52
+  },
+  "bor": {
+    "'": 47,
+    "0": 38,
+    "1": 45,
+    "2": 42,
+    "3": 50,
+    "4": 41,
+    "5": 48,
+    "6": 46,
+    "7": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "c": 25,
+    "d": 11,
+    "e": 5,
+    "f": 28,
+    "g": 12,
+    "h": 32,
+    "i": 10,
+    "j": 15,
+    "k": 16,
+    "l": 23,
+    "m": 13,
+    "n": 20,
+    "o": 8,
+    "p": 18,
+    "q": 31,
+    "r": 9,
+    "s": 21,
+    "t": 14,
+    "u": 7,
+    "v": 29,
+    "w": 19,
+    "x": 40,
+    "z": 33,
+    "|": 4,
+    "á": 30,
+    "â": 43,
+    "ã": 27,
+    "ç": 49,
+    "é": 26,
+    "ê": 39,
+    "í": 35,
+    "ó": 34,
+    "ô": 36,
+    "ú": 37,
+    "ü": 44,
+    "č": 22,
+    "–": 24
+  },
+  "bos": {
+    "!": 53,
+    "$": 59,
+    "%": 56,
+    "&": 69,
+    "'": 55,
+    "+": 62,
+    ",": 48,
+    "-": 36,
+    ".": 31,
+    "/": 47,
+    "0": 33,
+    "1": 34,
+    "2": 35,
+    "3": 42,
+    "4": 41,
+    "5": 39,
+    "6": 44,
+    "7": 45,
+    "8": 43,
+    "9": 37,
+    ":": 46,
+    ";": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 52,
+    "[": 74,
+    "]": 75,
+    "a": 5,
+    "b": 23,
+    "c": 25,
+    "d": 17,
+    "e": 8,
+    "f": 30,
+    "g": 22,
+    "h": 27,
+    "i": 6,
+    "j": 10,
+    "k": 15,
+    "l": 16,
+    "m": 19,
+    "n": 9,
+    "o": 7,
+    "p": 20,
+    "q": 50,
+    "r": 11,
+    "s": 13,
+    "t": 12,
+    "u": 14,
+    "v": 18,
+    "w": 40,
+    "x": 49,
+    "y": 38,
+    "z": 21,
+    "|": 4,
+    "°": 67,
+    "²": 73,
+    "á": 58,
+    "ã": 71,
+    "ç": 70,
+    "é": 61,
+    "í": 60,
+    "ó": 68,
+    "õ": 63,
+    "ú": 72,
+    "ü": 57,
+    "ć": 29,
+    "č": 24,
+    "đ": 32,
+    "ł": 76,
+    "š": 26,
+    "ž": 28,
+    "–": 54,
+    "—": 64,
+    "‘": 65,
+    "’": 66
+  },
+  "bov": {
+    "'": 38,
+    "-": 36,
+    "1": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "d": 19,
+    "e": 11,
+    "f": 22,
+    "g": 28,
+    "h": 32,
+    "i": 10,
+    "k": 7,
+    "l": 15,
+    "m": 13,
+    "n": 9,
+    "o": 18,
+    "p": 23,
+    "r": 33,
+    "s": 21,
+    "t": 14,
+    "u": 20,
+    "v": 30,
+    "w": 24,
+    "y": 17,
+    "z": 29,
+    "|": 4,
+    "á": 31,
+    "ã": 27,
+    "í": 35,
+    "ó": 37,
+    "ĩ": 16,
+    "ũ": 25,
+    "ɔ": 8,
+    "ɛ": 6,
+    "́": 34,
+    "̃": 26
+  },
+  "box": {
+    "-": 41,
+    "6": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 12,
+    "c": 46,
+    "d": 30,
+    "e": 7,
+    "f": 44,
+    "h": 19,
+    "i": 9,
+    "k": 20,
+    "l": 11,
+    "m": 14,
+    "n": 5,
+    "o": 8,
+    "p": 36,
+    "r": 25,
+    "s": 26,
+    "t": 32,
+    "u": 22,
+    "v": 47,
+    "w": 13,
+    "y": 18,
+    "z": 34,
+    "|": 4,
+    "à": 16,
+    "á": 10,
+    "ã": 33,
+    "è": 35,
+    "é": 24,
+    "ì": 17,
+    "í": 15,
+    "ò": 43,
+    "ó": 23,
+    "ù": 42,
+    "ú": 28,
+    "ĩ": 29,
+    "ń": 49,
+    "ũ": 45,
+    "ǹ": 50,
+    "ɓ": 21,
+    "ɔ": 37,
+    "ɛ": 27,
+    "ɲ": 40,
+    "̀": 38,
+    "́": 31,
+    "̃": 48,
+    "ṹ": 39
+  },
+  "bpr": {
+    "'": 36,
+    "-": 31,
+    "0": 33,
+    "1": 34,
+    "2": 35,
+    "3": 38,
+    "4": 37,
+    "5": 42,
+    "7": 39,
+    "8": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 41,
+    "d": 7,
+    "e": 14,
+    "f": 19,
+    "g": 13,
+    "h": 25,
+    "i": 8,
+    "k": 15,
+    "l": 10,
+    "m": 12,
+    "n": 6,
+    "o": 16,
+    "p": 30,
+    "r": 27,
+    "s": 17,
+    "t": 11,
+    "u": 9,
+    "w": 22,
+    "y": 18,
+    "|": 4,
+    "à": 23,
+    "è": 24,
+    "é": 21,
+    "ê": 29,
+    "ì": 32,
+    "ò": 26,
+    "ù": 28
+  },
+  "bps": {
+    "'": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "d": 7,
+    "e": 14,
+    "f": 17,
+    "g": 12,
+    "h": 26,
+    "i": 8,
+    "k": 15,
+    "l": 10,
+    "m": 13,
+    "n": 6,
+    "o": 18,
+    "p": 32,
+    "r": 27,
+    "s": 16,
+    "t": 11,
+    "u": 9,
+    "w": 21,
+    "y": 19,
+    "|": 4,
+    "à": 22,
+    "è": 24,
+    "é": 23,
+    "ê": 29,
+    "ì": 31,
+    "ò": 25,
+    "ù": 28
+  },
+  "bqc": {
+    "'": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 24,
+    "e": 21,
+    "f": 37,
+    "g": 18,
+    "h": 48,
+    "i": 8,
+    "k": 11,
+    "l": 10,
+    "m": 19,
+    "n": 16,
+    "o": 23,
+    "p": 17,
+    "s": 15,
+    "t": 32,
+    "u": 12,
+    "v": 38,
+    "w": 22,
+    "y": 27,
+    "z": 29,
+    "|": 4,
+    "à": 20,
+    "á": 26,
+    "ã": 25,
+    "è": 43,
+    "é": 46,
+    "ì": 34,
+    "í": 35,
+    "ò": 36,
+    "ó": 39,
+    "ù": 44,
+    "ú": 42,
+    "ĩ": 31,
+    "ń": 30,
+    "ũ": 33,
+    "ǹ": 47,
+    "ɔ": 40,
+    "ɛ": 6,
+    "̀": 28,
+    "́": 9,
+    "̃": 14,
+    "ḿ": 49,
+    "ṹ": 45,
+    "ↄ": 7
+  },
+  "bqi": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ء": 43,
+    "آ": 32,
+    "ئ": 41,
+    "ا": 6,
+    "ب": 14,
+    "ت": 16,
+    "ث": 39,
+    "ج": 28,
+    "ح": 26,
+    "خ": 19,
+    "د": 18,
+    "ذ": 37,
+    "ر": 11,
+    "ز": 20,
+    "س": 12,
+    "ش": 22,
+    "ص": 35,
+    "ض": 38,
+    "ط": 33,
+    "ظ": 40,
+    "ع": 25,
+    "غ": 34,
+    "ف": 31,
+    "ق": 29,
+    "ل": 21,
+    "م": 10,
+    "ن": 9,
+    "ه": 8,
+    "و": 7,
+    "ٚ": 17,
+    "ٛ": 36,
+    "پ": 24,
+    "چ": 30,
+    "ژ": 42,
+    "ک": 13,
+    "گ": 23,
+    "ۆ": 44,
+    "ی": 5,
+    "ێ": 45,
+    "ۨ": 27,
+    "ݚ": 15
+  },
+  "bqj": {
+    "'": 28,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 36,
+    "d": 34,
+    "e": 8,
+    "f": 19,
+    "g": 13,
+    "h": 21,
+    "i": 9,
+    "j": 17,
+    "k": 26,
+    "l": 14,
+    "m": 10,
+    "n": 6,
+    "o": 11,
+    "p": 37,
+    "r": 16,
+    "s": 18,
+    "t": 15,
+    "u": 7,
+    "v": 35,
+    "w": 23,
+    "y": 20,
+    "|": 4,
+    "á": 22,
+    "é": 30,
+    "í": 32,
+    "ñ": 29,
+    "ó": 33,
+    "ú": 25,
+    "ĉ": 31,
+    "ŋ": 27,
+    "ᵽ": 24,
+    "–": 39
+  },
+  "bqp": {
+    "'": 45,
+    "-": 52,
+    "0": 59,
+    "1": 50,
+    "2": 51,
+    "3": 53,
+    "4": 57,
+    "5": 56,
+    "6": 55,
+    "7": 54,
+    "8": 58,
+    "9": 60,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "d": 11,
+    "e": 23,
+    "f": 39,
+    "g": 14,
+    "h": 47,
+    "i": 9,
+    "k": 7,
+    "l": 24,
+    "m": 21,
+    "n": 6,
+    "o": 20,
+    "p": 22,
+    "r": 13,
+    "s": 17,
+    "t": 25,
+    "u": 15,
+    "v": 40,
+    "w": 35,
+    "y": 27,
+    "z": 34,
+    "|": 4,
+    "à": 19,
+    "á": 28,
+    "ã": 26,
+    "è": 43,
+    "é": 48,
+    "ì": 32,
+    "í": 36,
+    "ò": 29,
+    "ó": 37,
+    "ù": 42,
+    "ú": 41,
+    "ĩ": 33,
+    "ń": 38,
+    "ũ": 12,
+    "ǹ": 44,
+    "ɔ": 49,
+    "ɛ": 8,
+    "̀": 30,
+    "́": 31,
+    "̃": 18,
+    "ḿ": 46,
+    "ↄ": 10
+  },
+  "bre": {
+    "'": 21,
+    "-": 27,
+    "1": 38,
+    "2": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "c": 24,
+    "d": 15,
+    "e": 5,
+    "f": 30,
+    "g": 20,
+    "h": 11,
+    "i": 12,
+    "j": 31,
+    "k": 19,
+    "l": 14,
+    "m": 17,
+    "n": 7,
+    "o": 8,
+    "p": 23,
+    "q": 40,
+    "r": 9,
+    "s": 18,
+    "t": 10,
+    "u": 13,
+    "v": 22,
+    "w": 28,
+    "y": 33,
+    "z": 16,
+    "|": 4,
+    "à": 37,
+    "â": 35,
+    "è": 41,
+    "é": 36,
+    "ê": 34,
+    "ñ": 26,
+    "ù": 29,
+    "û": 42,
+    "ʼ": 32
+  },
+  "bru": {
+    "'": 50,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "c": 8,
+    "d": 35,
+    "e": 36,
+    "g": 12,
+    "h": 10,
+    "i": 7,
+    "k": 25,
+    "l": 19,
+    "m": 13,
+    "n": 6,
+    "o": 15,
+    "p": 20,
+    "q": 11,
+    "r": 18,
+    "s": 14,
+    "t": 9,
+    "u": 16,
+    "v": 44,
+    "y": 21,
+    "|": 4,
+    "á": 17,
+    "â": 22,
+    "é": 47,
+    "ê": 33,
+    "í": 38,
+    "ó": 42,
+    "ô": 43,
+    "ú": 49,
+    "ĩ": 30,
+    "ō": 46,
+    "ũ": 29,
+    "ơ": 37,
+    "ư": 48,
+    "ṓ": 39,
+    "ế": 45,
+    "ễ": 26,
+    "ố": 41,
+    "ỗ": 32,
+    "ớ": 23,
+    "ỡ": 24,
+    "ứ": 27,
+    "ữ": 31,
+    "‐": 34
+  },
+  "bsc": {
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 27,
+    "d": 13,
+    "e": 8,
+    "f": 26,
+    "g": 16,
+    "h": 34,
+    "i": 11,
+    "j": 35,
+    "k": 10,
+    "l": 18,
+    "m": 17,
+    "n": 7,
+    "o": 9,
+    "p": 30,
+    "r": 12,
+    "s": 24,
+    "t": 21,
+    "u": 23,
+    "w": 19,
+    "x": 15,
+    "y": 22,
+    "|": 4,
+    "ë": 6,
+    "ñ": 36,
+    "ŋ": 20,
+    "ŝ": 29,
+    "ƴ": 31,
+    "ɓ": 14,
+    "ɗ": 32,
+    "̃": 28,
+    "ỹ": 33
+  },
+  "bsq": {
+    "'": 62,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 15,
+    "b": 34,
+    "c": 43,
+    "d": 22,
+    "e": 37,
+    "f": 51,
+    "g": 27,
+    "h": 42,
+    "i": 30,
+    "j": 39,
+    "k": 10,
+    "m": 12,
+    "n": 7,
+    "o": 35,
+    "p": 25,
+    "s": 31,
+    "t": 41,
+    "u": 28,
+    "v": 55,
+    "w": 19,
+    "x": 52,
+    "y": 13,
+    "z": 40,
+    "|": 4,
+    "à": 16,
+    "á": 26,
+    "ã": 53,
+    "è": 23,
+    "é": 14,
+    "ì": 18,
+    "í": 20,
+    "ò": 33,
+    "ó": 38,
+    "ù": 36,
+    "ú": 32,
+    "ē": 61,
+    "ě": 24,
+    "ĩ": 48,
+    "ń": 54,
+    "ũ": 56,
+    "ū": 58,
+    "ǎ": 44,
+    "ǐ": 47,
+    "ǒ": 50,
+    "ǔ": 46,
+    "ǹ": 60,
+    "ɓ": 11,
+    "ɔ": 5,
+    "ɖ": 9,
+    "ɛ": 6,
+    "̀": 8,
+    "́": 17,
+    "̃": 29,
+    "̄": 57,
+    "̌": 49,
+    "ḿ": 45,
+    "ṹ": 59
+  },
+  "bss": {
+    "'": 60,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 13,
+    "b": 6,
+    "c": 32,
+    "d": 12,
+    "e": 9,
+    "f": 58,
+    "g": 21,
+    "h": 23,
+    "i": 34,
+    "j": 54,
+    "k": 19,
+    "l": 16,
+    "m": 10,
+    "n": 5,
+    "o": 26,
+    "p": 29,
+    "r": 45,
+    "s": 18,
+    "t": 20,
+    "u": 37,
+    "v": 61,
+    "w": 24,
+    "y": 15,
+    "z": 46,
+    "|": 4,
+    "á": 11,
+    "â": 39,
+    "é": 7,
+    "ê": 36,
+    "í": 30,
+    "î": 55,
+    "ó": 33,
+    "ô": 53,
+    "ú": 42,
+    "û": 57,
+    "ā": 49,
+    "ē": 35,
+    "ě": 41,
+    "ī": 51,
+    "ń": 47,
+    "ŋ": 25,
+    "ō": 52,
+    "ū": 56,
+    "ǎ": 48,
+    "ǐ": 50,
+    "ǒ": 38,
+    "ǔ": 59,
+    "ɔ": 17,
+    "ə": 22,
+    "ɛ": 8,
+    "ʼ": 27,
+    "́": 14,
+    "̂": 43,
+    "̄": 28,
+    "̌": 44,
+    "ḿ": 40
+  },
+  "btd": {
+    "'": 27,
+    "-": 24,
+    "0": 38,
+    "1": 32,
+    "2": 31,
+    "3": 33,
+    "5": 37,
+    "8": 36,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 25,
+    "d": 16,
+    "e": 8,
+    "f": 29,
+    "g": 10,
+    "h": 22,
+    "i": 7,
+    "j": 23,
+    "k": 9,
+    "l": 19,
+    "m": 11,
+    "n": 6,
+    "o": 12,
+    "p": 21,
+    "q": 40,
+    "r": 17,
+    "s": 13,
+    "t": 15,
+    "u": 14,
+    "w": 30,
+    "y": 26,
+    "z": 28,
+    "|": 4,
+    "à": 34,
+    "è": 18,
+    "ì": 39
+  },
+  "bts": {
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 29,
+    "d": 17,
+    "e": 19,
+    "f": 26,
+    "g": 13,
+    "h": 9,
+    "i": 7,
+    "j": 22,
+    "k": 21,
+    "l": 20,
+    "m": 10,
+    "n": 6,
+    "o": 8,
+    "p": 18,
+    "r": 14,
+    "s": 11,
+    "t": 15,
+    "u": 12,
+    "v": 27,
+    "w": 25,
+    "x": 30,
+    "y": 24,
+    "z": 28,
+    "|": 5
+  },
+  "btt": {
+    "'": 21,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 10,
+    "c": 53,
+    "d": 20,
+    "e": 5,
+    "f": 26,
+    "g": 11,
+    "h": 14,
+    "i": 7,
+    "j": 30,
+    "k": 16,
+    "l": 12,
+    "m": 13,
+    "n": 8,
+    "o": 17,
+    "p": 25,
+    "r": 27,
+    "s": 19,
+    "t": 22,
+    "u": 9,
+    "w": 18,
+    "y": 15,
+    "z": 34,
+    "|": 4,
+    "à": 28,
+    "á": 31,
+    "â": 46,
+    "è": 23,
+    "é": 44,
+    "ê": 52,
+    "ì": 29,
+    "í": 43,
+    "î": 45,
+    "ò": 37,
+    "ó": 49,
+    "ù": 32,
+    "ú": 41,
+    "û": 51,
+    "ā": 33,
+    "ē": 38,
+    "ě": 40,
+    "ī": 36,
+    "ı": 35,
+    "ō": 39,
+    "ū": 42,
+    "ǎ": 48,
+    "ǐ": 50,
+    "ǒ": 47,
+    "ǔ": 54
+  },
+  "btx": {
+    "'": 26,
+    "-": 23,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 34,
+    "4": 31,
+    "5": 35,
+    "6": 32,
+    "7": 37,
+    "8": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 24,
+    "d": 18,
+    "e": 7,
+    "g": 14,
+    "h": 20,
+    "i": 8,
+    "j": 21,
+    "k": 9,
+    "l": 15,
+    "m": 16,
+    "n": 6,
+    "o": 22,
+    "p": 19,
+    "r": 12,
+    "s": 11,
+    "t": 13,
+    "u": 10,
+    "w": 25,
+    "y": 27,
+    "z": 33,
+    "|": 4
+  },
+  "bud": {
+    "-": 22,
+    "0": 53,
+    "1": 52,
+    "2": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 10,
+    "c": 31,
+    "d": 16,
+    "e": 9,
+    "f": 28,
+    "g": 35,
+    "i": 5,
+    "j": 33,
+    "k": 8,
+    "l": 20,
+    "m": 14,
+    "n": 6,
+    "o": 24,
+    "p": 18,
+    "r": 36,
+    "s": 23,
+    "t": 12,
+    "u": 11,
+    "w": 42,
+    "y": 17,
+    "|": 4,
+    "à": 27,
+    "á": 19,
+    "è": 44,
+    "é": 40,
+    "ì": 25,
+    "í": 13,
+    "ï": 51,
+    "ò": 41,
+    "ù": 46,
+    "ú": 34,
+    "ā": 30,
+    "ē": 45,
+    "ī": 26,
+    "ĺ": 48,
+    "ń": 32,
+    "ŋ": 21,
+    "ū": 39,
+    "ǹ": 29,
+    "ɔ": 15,
+    "̀": 47,
+    "́": 38,
+    "̃": 43,
+    "̄": 37,
+    "ḿ": 49
+  },
+  "bul": {
+    "!": 74,
+    "\"": 75,
+    "%": 78,
+    "'": 79,
+    "+": 84,
+    ",": 67,
+    "-": 33,
+    ".": 35,
+    "/": 66,
+    "0": 34,
+    "1": 36,
+    "2": 38,
+    "3": 48,
+    "4": 46,
+    "5": 44,
+    "6": 49,
+    "7": 54,
+    "8": 51,
+    "9": 41,
+    ":": 55,
+    ";": 68,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 76,
+    "[": 86,
+    "]": 87,
+    "`": 93,
+    "a": 39,
+    "b": 69,
+    "c": 53,
+    "d": 62,
+    "e": 43,
+    "f": 73,
+    "g": 59,
+    "h": 61,
+    "i": 47,
+    "j": 83,
+    "k": 65,
+    "l": 57,
+    "m": 58,
+    "n": 45,
+    "o": 42,
+    "p": 56,
+    "q": 80,
+    "r": 40,
+    "s": 52,
+    "t": 50,
+    "u": 60,
+    "v": 70,
+    "w": 72,
+    "x": 82,
+    "y": 71,
+    "z": 81,
+    "|": 4,
+    "°": 85,
+    "²": 91,
+    "á": 88,
+    "õ": 89,
+    "̀": 94,
+    "а": 5,
+    "б": 23,
+    "в": 13,
+    "г": 22,
+    "д": 15,
+    "е": 8,
+    "ж": 26,
+    "з": 18,
+    "и": 6,
+    "й": 29,
+    "к": 14,
+    "л": 16,
+    "м": 19,
+    "н": 10,
+    "о": 7,
+    "п": 17,
+    "р": 11,
+    "с": 12,
+    "т": 9,
+    "у": 24,
+    "ф": 32,
+    "х": 30,
+    "ц": 27,
+    "ч": 25,
+    "ш": 31,
+    "щ": 28,
+    "ъ": 21,
+    "ь": 63,
+    "ю": 37,
+    "я": 20,
+    "ѐ": 95,
+    "ѝ": 77,
+    "–": 64,
+    "—": 92,
+    "№": 90
+  },
+  "bus": {
+    "'": 46,
+    "-": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 16,
+    "e": 11,
+    "f": 39,
+    "g": 15,
+    "h": 48,
+    "i": 10,
+    "k": 9,
+    "l": 22,
+    "m": 20,
+    "n": 7,
+    "o": 19,
+    "p": 23,
+    "r": 24,
+    "s": 21,
+    "t": 29,
+    "u": 14,
+    "v": 41,
+    "w": 27,
+    "y": 26,
+    "z": 28,
+    "|": 4,
+    "à": 17,
+    "á": 32,
+    "ã": 25,
+    "è": 31,
+    "é": 35,
+    "ì": 34,
+    "í": 36,
+    "ò": 43,
+    "ó": 40,
+    "ù": 45,
+    "ú": 42,
+    "ĩ": 33,
+    "ń": 37,
+    "ũ": 38,
+    "ǹ": 44,
+    "ɔ": 49,
+    "ɛ": 6,
+    "̀": 18,
+    "́": 30,
+    "̃": 12,
+    "ḿ": 47,
+    "ↄ": 8
+  },
+  "bvc": {
+    "'": 14,
+    "0": 27,
+    "1": 31,
+    "4": 28,
+    "5": 29,
+    "6": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 19,
+    "e": 9,
+    "f": 18,
+    "g": 20,
+    "h": 22,
+    "i": 6,
+    "j": 24,
+    "k": 12,
+    "l": 10,
+    "m": 13,
+    "n": 7,
+    "o": 8,
+    "p": 25,
+    "r": 17,
+    "s": 16,
+    "t": 15,
+    "u": 11,
+    "v": 26,
+    "w": 23,
+    "|": 4
+  },
+  "bvz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 29,
+    "d": 11,
+    "e": 6,
+    "f": 22,
+    "g": 20,
+    "h": 13,
+    "i": 8,
+    "j": 27,
+    "k": 16,
+    "l": 12,
+    "m": 7,
+    "n": 19,
+    "o": 9,
+    "p": 26,
+    "r": 25,
+    "s": 18,
+    "t": 15,
+    "u": 14,
+    "v": 17,
+    "w": 28,
+    "y": 24,
+    "z": 23,
+    "|": 4,
+    "à": 21
+  },
+  "bwq": {
+    "'": 27,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 21,
+    "e": 8,
+    "f": 25,
+    "g": 19,
+    "h": 28,
+    "i": 9,
+    "k": 13,
+    "l": 23,
+    "m": 11,
+    "n": 6,
+    "o": 17,
+    "p": 24,
+    "r": 15,
+    "s": 20,
+    "t": 16,
+    "u": 22,
+    "v": 29,
+    "w": 14,
+    "y": 12,
+    "z": 26,
+    "|": 4,
+    "à": 31,
+    "ì": 32,
+    "ɔ": 7,
+    "ɛ": 10,
+    "́": 30
+  },
+  "bwu": {
+    "'": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 25,
+    "d": 19,
+    "e": 7,
+    "f": 27,
+    "g": 20,
+    "h": 24,
+    "i": 6,
+    "j": 23,
+    "k": 11,
+    "l": 10,
+    "m": 9,
+    "n": 8,
+    "o": 18,
+    "p": 22,
+    "r": 21,
+    "s": 17,
+    "t": 15,
+    "u": 13,
+    "v": 28,
+    "w": 16,
+    "y": 12,
+    "z": 26,
+    "|": 4
+  },
+  "byr": {
+    "'": 12,
+    "-": 33,
+    "0": 29,
+    "1": 30,
+    "2": 31,
+    "3": 37,
+    "4": 34,
+    "5": 35,
+    "6": 32,
+    "7": 38,
+    "8": 39,
+    "9": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "d": 17,
+    "e": 9,
+    "g": 16,
+    "h": 25,
+    "i": 15,
+    "j": 24,
+    "k": 14,
+    "l": 19,
+    "m": 13,
+    "n": 7,
+    "o": 28,
+    "p": 27,
+    "r": 10,
+    "s": 18,
+    "t": 23,
+    "u": 21,
+    "v": 20,
+    "w": 11,
+    "y": 8,
+    "z": 26,
+    "|": 6,
+    "ɨ": 5
+  },
+  "bzh": {
+    "'": 32,
+    "-": 30,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "3": 39,
+    "4": 37,
+    "5": 35,
+    "6": 41,
+    "7": 42,
+    "8": 43,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 23,
+    "e": 7,
+    "g": 14,
+    "h": 17,
+    "i": 8,
+    "j": 29,
+    "k": 25,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 9,
+    "p": 26,
+    "q": 36,
+    "r": 16,
+    "s": 12,
+    "t": 24,
+    "u": 11,
+    "v": 19,
+    "w": 28,
+    "y": 22,
+    "|": 4,
+    "ë": 10,
+    "ö": 27,
+    "ġ": 21,
+    "ḳ": 20,
+    "—": 40
+  },
+  "bzi": {
+    "'": 61,
+    "-": 65,
+    "0": 52,
+    "1": 55,
+    "2": 48,
+    "3": 57,
+    "4": 58,
+    "5": 62,
+    "6": 64,
+    "7": 59,
+    "8": 66,
+    "9": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "m": 72,
+    "q": 71,
+    "|": 4,
+    "ก": 18,
+    "ข": 51,
+    "ค": 24,
+    "ง": 6,
+    "จ": 19,
+    "ช": 23,
+    "ซ": 27,
+    "ญ": 40,
+    "ฎ": 68,
+    "ฐ": 54,
+    "ณ": 60,
+    "ด": 34,
+    "ต": 30,
+    "ถ": 49,
+    "ท": 35,
+    "ธ": 44,
+    "น": 10,
+    "บ": 22,
+    "ป": 25,
+    "ผ": 56,
+    "พ": 28,
+    "ฟ": 46,
+    "ภ": 50,
+    "ม": 13,
+    "ย": 11,
+    "ร": 33,
+    "ล": 16,
+    "ว": 17,
+    "ศ": 42,
+    "ษ": 43,
+    "ส": 37,
+    "ห": 45,
+    "ฬ": 73,
+    "อ": 8,
+    "ฮ": 26,
+    "ะ": 38,
+    "ั": 41,
+    "า": 5,
+    "ิ": 29,
+    "ี": 14,
+    "ึ": 70,
+    "ื": 21,
+    "ุ": 47,
+    "ู": 20,
+    "ฺ": 31,
+    "เ": 15,
+    "แ": 12,
+    "โ": 32,
+    "ไ": 36,
+    "็": 53,
+    "่": 7,
+    "้": 9,
+    "๋": 74,
+    "์": 39,
+    "ํ": 63,
+    "‍": 69
+  },
+  "bzj": {
+    "'": 30,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 28,
+    "d": 10,
+    "e": 7,
+    "f": 20,
+    "g": 22,
+    "h": 8,
+    "i": 9,
+    "j": 27,
+    "k": 18,
+    "l": 15,
+    "m": 19,
+    "n": 6,
+    "o": 11,
+    "p": 21,
+    "r": 17,
+    "s": 12,
+    "t": 14,
+    "u": 16,
+    "v": 26,
+    "w": 13,
+    "y": 23,
+    "z": 25,
+    "|": 4,
+    "—": 31
+  },
+  "caa": {
+    "'": 11,
+    "-": 45,
+    "0": 37,
+    "1": 38,
+    "2": 39,
+    "3": 43,
+    "4": 41,
+    "5": 40,
+    "6": 44,
+    "7": 47,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 14,
+    "d": 26,
+    "e": 6,
+    "f": 34,
+    "g": 28,
+    "h": 21,
+    "i": 10,
+    "j": 19,
+    "k": 42,
+    "l": 25,
+    "m": 15,
+    "n": 9,
+    "o": 13,
+    "p": 22,
+    "q": 23,
+    "r": 12,
+    "s": 18,
+    "t": 8,
+    "u": 7,
+    "v": 35,
+    "w": 24,
+    "x": 20,
+    "y": 16,
+    "z": 27,
+    "|": 4,
+    "á": 30,
+    "é": 33,
+    "í": 31,
+    "ó": 36,
+    "ú": 29,
+    "—": 32
+  },
+  "cab": {
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 30,
+    "d": 15,
+    "e": 11,
+    "f": 24,
+    "g": 10,
+    "h": 14,
+    "i": 7,
+    "k": 27,
+    "l": 9,
+    "m": 16,
+    "n": 8,
+    "o": 25,
+    "p": 32,
+    "r": 12,
+    "s": 17,
+    "t": 19,
+    "u": 6,
+    "w": 23,
+    "y": 29,
+    "|": 4,
+    "á": 21,
+    "é": 26,
+    "í": 22,
+    "ñ": 20,
+    "ó": 31,
+    "ú": 28,
+    "ü": 18,
+    "ǘ": 34,
+    "—": 33
+  },
+  "cac-dialect_sanmateoixtatan": {
+    "'": 6,
+    "-": 29,
+    "0": 37,
+    "1": 40,
+    "2": 39,
+    "3": 41,
+    "4": 38,
+    "5": 42,
+    "6": 43,
+    "7": 45,
+    "8": 46,
+    "9": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 7,
+    "d": 21,
+    "e": 11,
+    "f": 35,
+    "g": 30,
+    "h": 18,
+    "i": 8,
+    "j": 14,
+    "l": 12,
+    "m": 24,
+    "n": 9,
+    "o": 10,
+    "p": 25,
+    "q": 28,
+    "r": 27,
+    "s": 16,
+    "t": 13,
+    "u": 17,
+    "v": 23,
+    "x": 20,
+    "y": 15,
+    "z": 26,
+    "|": 4,
+    "á": 34,
+    "é": 33,
+    "í": 32,
+    "ñ": 47,
+    "ó": 36,
+    "ú": 31,
+    "̈": 22
+  },
+  "cac-dialect_sansebastiancoatan": {
+    "'": 6,
+    "-": 38,
+    "0": 41,
+    "1": 42,
+    "2": 40,
+    "3": 44,
+    "4": 39,
+    "5": 47,
+    "6": 45,
+    "7": 46,
+    "8": 48,
+    "9": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 7,
+    "d": 28,
+    "e": 11,
+    "f": 36,
+    "g": 30,
+    "h": 20,
+    "i": 9,
+    "j": 13,
+    "k": 49,
+    "l": 14,
+    "m": 22,
+    "n": 8,
+    "o": 12,
+    "p": 25,
+    "q": 26,
+    "r": 27,
+    "s": 17,
+    "t": 10,
+    "u": 16,
+    "v": 34,
+    "w": 21,
+    "x": 18,
+    "y": 15,
+    "z": 23,
+    "|": 4,
+    "á": 35,
+    "é": 33,
+    "í": 32,
+    "ó": 37,
+    "ú": 29,
+    "̈": 24,
+    "—": 31
+  },
+  "cak-dialect_central": {
+    "'": 8,
+    "-": 27,
+    "0": 38,
+    "1": 40,
+    "2": 41,
+    "3": 45,
+    "4": 39,
+    "5": 42,
+    "6": 43,
+    "7": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 24,
+    "c": 14,
+    "d": 29,
+    "e": 11,
+    "f": 35,
+    "g": 33,
+    "h": 15,
+    "i": 5,
+    "j": 12,
+    "k": 10,
+    "l": 18,
+    "m": 22,
+    "n": 9,
+    "o": 13,
+    "p": 26,
+    "q": 19,
+    "r": 7,
+    "s": 25,
+    "t": 17,
+    "u": 16,
+    "v": 36,
+    "w": 23,
+    "x": 20,
+    "y": 21,
+    "z": 28,
+    "|": 4,
+    "à": 46,
+    "á": 30,
+    "é": 34,
+    "ì": 47,
+    "í": 32,
+    "ó": 37,
+    "ú": 31
+  },
+  "cak-dialect_santamariadejesus": {
+    "'": 7,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 27,
+    "c": 10,
+    "d": 28,
+    "e": 12,
+    "f": 30,
+    "g": 32,
+    "h": 16,
+    "i": 5,
+    "j": 14,
+    "k": 21,
+    "l": 20,
+    "m": 19,
+    "n": 9,
+    "o": 13,
+    "p": 24,
+    "q": 23,
+    "r": 8,
+    "s": 18,
+    "t": 15,
+    "u": 11,
+    "v": 25,
+    "x": 17,
+    "y": 22,
+    "z": 29,
+    "|": 4,
+    "á": 31,
+    "é": 35,
+    "í": 33,
+    "ñ": 37,
+    "ó": 36,
+    "ú": 34
+  },
+  "cak-dialect_santodomingoxenacoj": {
+    "-": 31,
+    "0": 38,
+    "1": 39,
+    "2": 41,
+    "3": 42,
+    "4": 40,
+    "5": 43,
+    "6": 44,
+    "7": 46,
+    "8": 47,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 26,
+    "c": 15,
+    "d": 24,
+    "e": 10,
+    "f": 35,
+    "g": 29,
+    "h": 16,
+    "i": 5,
+    "j": 13,
+    "k": 11,
+    "l": 19,
+    "m": 22,
+    "n": 9,
+    "o": 14,
+    "p": 25,
+    "q": 18,
+    "r": 7,
+    "s": 21,
+    "t": 17,
+    "u": 12,
+    "v": 37,
+    "w": 27,
+    "x": 20,
+    "y": 23,
+    "z": 28,
+    "|": 4,
+    "á": 30,
+    "é": 34,
+    "í": 33,
+    "ñ": 48,
+    "ó": 36,
+    "ú": 32,
+    "ꞌ": 8
+  },
+  "cak-dialect_southcentral": {
+    "'": 11,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 27,
+    "c": 9,
+    "d": 29,
+    "e": 12,
+    "f": 31,
+    "g": 32,
+    "h": 15,
+    "i": 5,
+    "j": 14,
+    "k": 22,
+    "l": 17,
+    "m": 18,
+    "n": 8,
+    "o": 13,
+    "p": 24,
+    "q": 21,
+    "r": 7,
+    "s": 20,
+    "t": 16,
+    "u": 10,
+    "v": 25,
+    "x": 19,
+    "y": 23,
+    "z": 28,
+    "|": 4,
+    "á": 30,
+    "é": 35,
+    "í": 34,
+    "ñ": 37,
+    "ó": 36,
+    "ú": 33
+  },
+  "cak-dialect_western": {
+    "'": 9,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 24,
+    "c": 7,
+    "d": 28,
+    "e": 10,
+    "f": 29,
+    "g": 33,
+    "h": 13,
+    "i": 5,
+    "j": 14,
+    "k": 20,
+    "l": 18,
+    "m": 21,
+    "n": 12,
+    "o": 15,
+    "p": 26,
+    "q": 22,
+    "r": 8,
+    "s": 23,
+    "t": 16,
+    "u": 11,
+    "v": 36,
+    "w": 25,
+    "x": 19,
+    "y": 17,
+    "z": 27,
+    "|": 4,
+    "á": 30,
+    "é": 35,
+    "í": 32,
+    "ñ": 38,
+    "ó": 34,
+    "ú": 31
+  },
+  "cak-dialect_yepocapa": {
+    "'": 11,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 9,
+    "d": 29,
+    "e": 6,
+    "f": 30,
+    "g": 31,
+    "h": 15,
+    "i": 7,
+    "j": 14,
+    "k": 21,
+    "l": 22,
+    "m": 18,
+    "n": 8,
+    "o": 13,
+    "p": 26,
+    "q": 23,
+    "r": 10,
+    "s": 20,
+    "t": 16,
+    "u": 12,
+    "v": 27,
+    "x": 19,
+    "y": 17,
+    "z": 28,
+    "|": 4
+  },
+  "cap": {
+    "'": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 36,
+    "c": 18,
+    "d": 33,
+    "e": 17,
+    "f": 37,
+    "g": 41,
+    "h": 8,
+    "i": 6,
+    "j": 28,
+    "k": 9,
+    "l": 14,
+    "m": 21,
+    "n": 7,
+    "o": 23,
+    "p": 19,
+    "q": 16,
+    "r": 22,
+    "s": 20,
+    "t": 10,
+    "u": 13,
+    "v": 44,
+    "w": 24,
+    "x": 25,
+    "y": 27,
+    "z": 42,
+    "|": 5,
+    "á": 46,
+    "ä": 31,
+    "é": 40,
+    "ë": 35,
+    "í": 39,
+    "ï": 29,
+    "ñ": 26,
+    "ó": 45,
+    "ö": 30,
+    "ú": 43,
+    "ü": 38,
+    "ć": 15,
+    "ś": 11,
+    "ź": 12,
+    "–": 34
+  },
+  "car": {
+    "'": 10,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 21,
+    "d": 22,
+    "e": 13,
+    "f": 25,
+    "g": 23,
+    "h": 26,
+    "i": 15,
+    "j": 20,
+    "k": 12,
+    "l": 24,
+    "m": 8,
+    "n": 7,
+    "o": 5,
+    "p": 16,
+    "r": 9,
+    "s": 19,
+    "t": 14,
+    "u": 17,
+    "v": 28,
+    "w": 18,
+    "y": 11,
+    "z": 29,
+    "|": 4
+  },
+  "cas": {
+    "'": 5,
+    "-": 44,
+    "0": 56,
+    "1": 48,
+    "2": 50,
+    "3": 52,
+    "4": 51,
+    "5": 53,
+    "6": 57,
+    "7": 58,
+    "8": 55,
+    "9": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 25,
+    "c": 15,
+    "d": 17,
+    "e": 9,
+    "f": 34,
+    "g": 45,
+    "h": 18,
+    "i": 6,
+    "j": 10,
+    "l": 38,
+    "m": 11,
+    "n": 13,
+    "o": 20,
+    "p": 23,
+    "q": 21,
+    "r": 19,
+    "s": 16,
+    "t": 14,
+    "u": 12,
+    "v": 27,
+    "x": 59,
+    "y": 7,
+    "z": 49,
+    "|": 4,
+    "á": 40,
+    "â": 60,
+    "ä": 22,
+    "é": 33,
+    "í": 41,
+    "ñ": 30,
+    "ó": 42,
+    "ú": 36,
+    "ÿ": 61,
+    "ĉ": 24,
+    "ŝ": 43,
+    "́": 37,
+    "̂": 32,
+    "̈": 47,
+    "̣": 35,
+    "ạ": 29,
+    "ẹ": 46,
+    "ị": 31,
+    "ọ": 28,
+    "ụ": 26,
+    "—": 39
+  },
+  "cat": {
+    "!": 64,
+    "\"": 58,
+    "%": 59,
+    "&": 73,
+    "'": 22,
+    "+": 76,
+    ",": 57,
+    "-": 40,
+    ".": 52,
+    "/": 56,
+    "0": 35,
+    "1": 37,
+    "2": 41,
+    "3": 50,
+    "4": 48,
+    "5": 47,
+    "6": 51,
+    "7": 54,
+    "8": 49,
+    "9": 45,
+    ":": 55,
+    ";": 63,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 77,
+    "[": 69,
+    "]": 70,
+    "a": 6,
+    "b": 21,
+    "c": 16,
+    "d": 14,
+    "e": 5,
+    "f": 23,
+    "g": 20,
+    "h": 25,
+    "i": 9,
+    "j": 32,
+    "k": 38,
+    "l": 11,
+    "m": 17,
+    "n": 10,
+    "o": 13,
+    "p": 18,
+    "q": 24,
+    "r": 8,
+    "s": 7,
+    "t": 12,
+    "u": 15,
+    "v": 19,
+    "w": 43,
+    "x": 27,
+    "y": 33,
+    "z": 36,
+    "|": 4,
+    "’": 68,
+    "“": 74,
+    "”": 66,
+    "–": 65,
+    "—": 60,
+    "¥": 67,
+    "°": 79,
+    "²": 62,
+    "·": 44,
+    "º": 71,
+    "à": 29,
+    "á": 61,
+    "ã": 78,
+    "ä": 101,
+    "å": 95,
+    "ç": 39,
+    "è": 31,
+    "é": 26,
+    "ë": 92,
+    "ì": 103,
+    "í": 30,
+    "ï": 46,
+    "ð": 105,
+    "ñ": 80,
+    "ò": 34,
+    "ó": 28,
+    "ô": 84,
+    "õ": 75,
+    "ö": 72,
+    "ø": 83,
+    "ù": 99,
+    "ú": 42,
+    "ü": 53,
+    "ā": 100,
+    "č": 88,
+    "ė": 96,
+    "ı": 89,
+    "ł": 93,
+    "ň": 97,
+    "ō": 98,
+    "ő": 102,
+    "ř": 85,
+    "ş": 90,
+    "š": 91,
+    "ţ": 94,
+    "ū": 86,
+    "ž": 87,
+    "ὑ": 104,
+    "–": 81,
+    "—": 82
+  },
+  "cax": {
+    "'": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 23,
+    "d": 27,
+    "e": 15,
+    "f": 33,
+    "g": 30,
+    "h": 20,
+    "i": 6,
+    "j": 24,
+    "k": 12,
+    "l": 25,
+    "m": 13,
+    "n": 7,
+    "o": 11,
+    "p": 17,
+    "q": 38,
+    "r": 18,
+    "s": 16,
+    "t": 8,
+    "u": 10,
+    "v": 36,
+    "x": 14,
+    "y": 21,
+    "z": 37,
+    "|": 5,
+    "á": 35,
+    "é": 31,
+    "í": 28,
+    "ñ": 22,
+    "ó": 34,
+    "ú": 26,
+    "ɨ": 9,
+    "–": 29
+  },
+  "cbc": {
+    "'": 45,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 10,
+    "d": 31,
+    "e": 12,
+    "f": 39,
+    "g": 26,
+    "h": 40,
+    "i": 7,
+    "j": 18,
+    "k": 44,
+    "l": 32,
+    "m": 14,
+    "n": 16,
+    "o": 11,
+    "p": 19,
+    "q": 27,
+    "r": 6,
+    "s": 22,
+    "t": 15,
+    "u": 8,
+    "v": 42,
+    "w": 30,
+    "x": 43,
+    "y": 23,
+    "z": 41,
+    "|": 4,
+    "á": 29,
+    "ã": 13,
+    "é": 37,
+    "í": 34,
+    "ñ": 25,
+    "ó": 35,
+    "õ": 28,
+    "ú": 33,
+    "ĩ": 21,
+    "ũ": 17,
+    "̶": 9,
+    "ẽ": 24,
+    "—": 38
+  },
+  "cbi": {
+    "'": 18,
+    "-": 30,
+    "0": 34,
+    "1": 38,
+    "2": 40,
+    "3": 44,
+    "4": 41,
+    "5": 42,
+    "6": 45,
+    "7": 47,
+    "8": 48,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 22,
+    "d": 17,
+    "e": 8,
+    "f": 29,
+    "g": 26,
+    "h": 19,
+    "i": 9,
+    "j": 16,
+    "k": 15,
+    "l": 12,
+    "m": 13,
+    "n": 7,
+    "o": 28,
+    "p": 23,
+    "q": 39,
+    "r": 21,
+    "s": 11,
+    "t": 10,
+    "u": 6,
+    "v": 25,
+    "w": 27,
+    "x": 43,
+    "y": 14,
+    "z": 37,
+    "|": 5,
+    "á": 36,
+    "é": 33,
+    "í": 32,
+    "ñ": 24,
+    "ó": 35,
+    "ú": 31,
+    "ü": 49
+  },
+  "cbr": {
+    "'": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 9,
+    "d": 25,
+    "e": 24,
+    "f": 35,
+    "g": 33,
+    "h": 21,
+    "i": 7,
+    "j": 26,
+    "k": 40,
+    "l": 30,
+    "m": 12,
+    "n": 6,
+    "o": 19,
+    "p": 18,
+    "q": 17,
+    "r": 16,
+    "s": 11,
+    "t": 13,
+    "u": 8,
+    "v": 36,
+    "x": 15,
+    "y": 39,
+    "z": 37,
+    "|": 4,
+    "á": 22,
+    "é": 31,
+    "ë": 10,
+    "í": 20,
+    "ñ": 23,
+    "ó": 32,
+    "ú": 27,
+    "ü": 38,
+    "́": 29,
+    "—": 28
+  },
+  "cbs": {
+    "'": 42,
+    "0": 31,
+    "1": 37,
+    "2": 38,
+    "3": 45,
+    "4": 41,
+    "5": 44,
+    "6": 46,
+    "7": 40,
+    "8": 48,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "c": 22,
+    "d": 16,
+    "e": 10,
+    "f": 32,
+    "g": 30,
+    "h": 19,
+    "i": 7,
+    "j": 13,
+    "k": 9,
+    "l": 26,
+    "m": 15,
+    "n": 6,
+    "o": 23,
+    "p": 20,
+    "q": 39,
+    "r": 24,
+    "s": 12,
+    "t": 14,
+    "u": 8,
+    "v": 34,
+    "w": 17,
+    "x": 21,
+    "y": 18,
+    "z": 36,
+    "|": 5,
+    "á": 33,
+    "é": 28,
+    "í": 29,
+    "ñ": 43,
+    "ó": 35,
+    "ú": 27,
+    "–": 25
+  },
+  "cbt": {
+    "'": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 16,
+    "d": 26,
+    "e": 28,
+    "f": 30,
+    "h": 11,
+    "i": 7,
+    "l": 27,
+    "m": 20,
+    "n": 5,
+    "o": 8,
+    "p": 14,
+    "q": 19,
+    "r": 15,
+    "s": 17,
+    "t": 10,
+    "u": 12,
+    "y": 18,
+    "|": 6,
+    "á": 22,
+    "ë": 9,
+    "í": 23,
+    "ó": 25,
+    "́": 29,
+    "—": 24,
+    "ꞌ": 13
+  },
+  "cbu": {
+    "'": 27,
+    "-": 43,
+    "0": 34,
+    "1": 35,
+    "2": 37,
+    "3": 42,
+    "4": 39,
+    "5": 40,
+    "6": 41,
+    "7": 44,
+    "8": 46,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 19,
+    "d": 25,
+    "e": 17,
+    "f": 33,
+    "g": 18,
+    "h": 12,
+    "i": 6,
+    "j": 28,
+    "k": 16,
+    "l": 20,
+    "m": 14,
+    "n": 7,
+    "o": 22,
+    "p": 13,
+    "r": 11,
+    "s": 8,
+    "t": 9,
+    "u": 10,
+    "v": 36,
+    "w": 21,
+    "x": 30,
+    "y": 15,
+    "z": 23,
+    "|": 5,
+    "á": 24,
+    "é": 38,
+    "í": 29,
+    "ó": 32,
+    "ú": 31
+  },
+  "cbv": {
+    "'": 38,
+    "1": 49,
+    "4": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 13,
+    "d": 15,
+    "e": 12,
+    "f": 40,
+    "g": 33,
+    "h": 7,
+    "i": 6,
+    "j": 10,
+    "k": 45,
+    "l": 34,
+    "m": 22,
+    "n": 8,
+    "o": 18,
+    "p": 16,
+    "q": 35,
+    "r": 31,
+    "s": 27,
+    "t": 9,
+    "u": 30,
+    "v": 41,
+    "w": 19,
+    "x": 44,
+    "y": 26,
+    "z": 42,
+    "|": 4,
+    "á": 14,
+    "ã": 20,
+    "ç": 47,
+    "é": 25,
+    "í": 21,
+    "ñ": 29,
+    "ò": 46,
+    "ó": 28,
+    "õ": 43,
+    "ú": 36,
+    "ĩ": 24,
+    "ʉ": 11,
+    "́": 23,
+    "̃": 32,
+    "ẽ": 37,
+    "—": 39
+  },
+  "cce": {
+    "'": 29,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 33,
+    "a": 5,
+    "b": 23,
+    "d": 20,
+    "e": 11,
+    "f": 27,
+    "g": 18,
+    "h": 14,
+    "i": 6,
+    "j": 28,
+    "k": 10,
+    "l": 17,
+    "m": 13,
+    "n": 7,
+    "o": 12,
+    "p": 26,
+    "r": 24,
+    "s": 16,
+    "t": 9,
+    "u": 8,
+    "v": 19,
+    "w": 15,
+    "x": 21,
+    "y": 22,
+    "z": 25,
+    "|": 4,
+    "é": 32,
+    "ñ": 30
+  },
+  "cco": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 20,
+    "d": 19,
+    "e": 8,
+    "f": 28,
+    "g": 15,
+    "h": 43,
+    "i": 7,
+    "j": 10,
+    "k": 49,
+    "l": 17,
+    "m": 24,
+    "n": 9,
+    "o": 14,
+    "p": 45,
+    "q": 31,
+    "r": 30,
+    "s": 18,
+    "t": 33,
+    "u": 16,
+    "v": 47,
+    "x": 48,
+    "y": 42,
+    "z": 46,
+    "|": 4,
+    "á": 26,
+    "é": 25,
+    "í": 27,
+    "ñ": 36,
+    "ó": 37,
+    "ø": 39,
+    "ú": 38,
+    "ü": 41,
+    "ǿ": 40,
+    "ɨ": 13,
+    "ʉ": 35,
+    "ˆ": 34,
+    "ˉ": 21,
+    "ˊ": 23,
+    "ˋ": 12,
+    "́": 29,
+    "̃": 22,
+    "̱": 11,
+    "—": 44
+  },
+  "cdj": {
+    "'": 66,
+    "-": 50,
+    "0": 68,
+    "3": 67,
+    "5": 65,
+    "7": 64,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 55,
+    "ं": 10,
+    "ः": 63,
+    "अ": 28,
+    "आ": 38,
+    "इ": 54,
+    "ई": 29,
+    "उ": 40,
+    "ऊ": 58,
+    "ए": 53,
+    "ऐ": 36,
+    "ओ": 62,
+    "औ": 59,
+    "क": 11,
+    "ख": 43,
+    "ग": 32,
+    "घ": 52,
+    "च": 39,
+    "छ": 46,
+    "ज": 21,
+    "झ": 41,
+    "ञ": 57,
+    "ट": 48,
+    "ठ": 47,
+    "ड": 31,
+    "ढ": 56,
+    "ण": 24,
+    "त": 9,
+    "थ": 42,
+    "द": 30,
+    "ध": 45,
+    "न": 14,
+    "प": 22,
+    "फ": 49,
+    "ब": 26,
+    "भ": 37,
+    "म": 17,
+    "य": 18,
+    "र": 6,
+    "ल": 25,
+    "व": 23,
+    "श": 27,
+    "ष": 51,
+    "स": 15,
+    "ह": 20,
+    "़": 35,
+    "ा": 5,
+    "ि": 8,
+    "ी": 13,
+    "ु": 19,
+    "ू": 33,
+    "ृ": 61,
+    "े": 16,
+    "ै": 7,
+    "ो": 34,
+    "ौ": 44,
+    "्": 12,
+    "‍": 60
+  },
+  "ceb": {
+    "!": 50,
+    "\"": 48,
+    "$": 51,
+    "%": 56,
+    "&": 60,
+    "'": 44,
+    "+": 59,
+    ",": 42,
+    "-": 24,
+    ".": 30,
+    "/": 47,
+    "0": 26,
+    "1": 29,
+    "2": 31,
+    "3": 38,
+    "4": 36,
+    "5": 34,
+    "6": 39,
+    "7": 41,
+    "8": 37,
+    "9": 33,
+    ":": 43,
+    ";": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 49,
+    "[": 65,
+    "]": 66,
+    "a": 4,
+    "b": 21,
+    "c": 25,
+    "d": 18,
+    "e": 17,
+    "f": 27,
+    "g": 7,
+    "h": 22,
+    "i": 8,
+    "j": 32,
+    "k": 14,
+    "l": 15,
+    "m": 11,
+    "n": 6,
+    "o": 10,
+    "p": 16,
+    "q": 45,
+    "r": 19,
+    "s": 9,
+    "t": 12,
+    "u": 13,
+    "v": 28,
+    "w": 23,
+    "x": 40,
+    "y": 20,
+    "z": 35,
+    "|": 5,
+    "£": 63,
+    "¥": 64,
+    "°": 61,
+    "²": 72,
+    "á": 55,
+    "â": 69,
+    "ã": 67,
+    "ç": 58,
+    "é": 73,
+    "í": 57,
+    "ó": 71,
+    "õ": 70,
+    "ú": 62,
+    "ü": 52,
+    "–": 54,
+    "—": 53,
+    "’": 68
+  },
+  "ceg": {
+    "'": 47,
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 26,
+    "c": 20,
+    "d": 25,
+    "e": 5,
+    "f": 37,
+    "g": 35,
+    "h": 7,
+    "i": 15,
+    "j": 18,
+    "k": 14,
+    "l": 21,
+    "m": 22,
+    "n": 17,
+    "o": 6,
+    "p": 19,
+    "q": 42,
+    "r": 8,
+    "s": 11,
+    "t": 12,
+    "u": 13,
+    "v": 41,
+    "w": 23,
+    "x": 45,
+    "y": 16,
+    "z": 27,
+    "|": 4,
+    "á": 40,
+    "ã": 31,
+    "é": 36,
+    "í": 38,
+    "ñ": 46,
+    "ó": 39,
+    "õ": 24,
+    "ú": 34,
+    "ĩ": 28,
+    "ũ": 30,
+    "ɨ": 9,
+    "̃": 43,
+    "̵": 29,
+    "ẽ": 32,
+    "—": 33
+  },
+  "cek": {
+    "'": 36,
+    "-": 22,
+    "0": 35,
+    "1": 38,
+    "2": 37,
+    "3": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 23,
+    "d": 25,
+    "e": 10,
+    "f": 34,
+    "g": 8,
+    "h": 12,
+    "i": 7,
+    "j": 30,
+    "k": 13,
+    "l": 14,
+    "m": 16,
+    "n": 6,
+    "o": 29,
+    "p": 21,
+    "q": 31,
+    "r": 26,
+    "s": 18,
+    "t": 11,
+    "u": 19,
+    "v": 24,
+    "w": 15,
+    "x": 32,
+    "y": 17,
+    "z": 33,
+    "|": 4,
+    "ä": 9,
+    "ö": 28,
+    "ü": 20
+  },
+  "ces": {
+    "!": 69,
+    "\"": 64,
+    "%": 62,
+    "&": 75,
+    "'": 63,
+    "+": 76,
+    ",": 60,
+    "-": 54,
+    ".": 45,
+    "/": 57,
+    "0": 38,
+    "1": 39,
+    "2": 41,
+    "3": 51,
+    "4": 47,
+    "5": 46,
+    "6": 49,
+    "7": 53,
+    "8": 50,
+    "9": 44,
+    ":": 58,
+    ";": 74,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 71,
+    "]": 72,
+    "a": 8,
+    "b": 27,
+    "c": 21,
+    "d": 16,
+    "e": 6,
+    "f": 37,
+    "g": 36,
+    "h": 22,
+    "i": 11,
+    "j": 26,
+    "k": 15,
+    "l": 13,
+    "m": 18,
+    "n": 7,
+    "o": 5,
+    "p": 17,
+    "q": 59,
+    "r": 14,
+    "s": 10,
+    "t": 9,
+    "u": 19,
+    "v": 12,
+    "w": 42,
+    "x": 43,
+    "y": 25,
+    "z": 23,
+    "|": 4,
+    "°": 65,
+    "²": 78,
+    "×": 77,
+    "á": 24,
+    "ã": 81,
+    "ä": 84,
+    "ç": 79,
+    "è": 83,
+    "é": 29,
+    "í": 20,
+    "ï": 85,
+    "ó": 52,
+    "õ": 66,
+    "ö": 68,
+    "ú": 40,
+    "ü": 73,
+    "ý": 32,
+    "ā": 80,
+    "č": 33,
+    "ď": 56,
+    "ě": 28,
+    "ł": 82,
+    "ň": 48,
+    "ř": 30,
+    "š": 34,
+    "ť": 55,
+    "ů": 35,
+    "ž": 31,
+    "–": 61,
+    "—": 70,
+    "⁠": 67
+  },
+  "cfm": {
+    "'": 33,
+    "-": 29,
+    "1": 36,
+    "2": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 19,
+    "d": 22,
+    "e": 14,
+    "f": 26,
+    "g": 15,
+    "h": 7,
+    "i": 8,
+    "j": 28,
+    "k": 12,
+    "l": 11,
+    "m": 13,
+    "n": 6,
+    "o": 21,
+    "p": 18,
+    "q": 32,
+    "r": 17,
+    "s": 16,
+    "t": 10,
+    "u": 9,
+    "v": 27,
+    "w": 20,
+    "x": 31,
+    "y": 30,
+    "z": 23,
+    "|": 4,
+    "­": 35,
+    "ò": 25
+  },
+  "cgc": {
+    "'": 31,
+    "-": 23,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 25,
+    "d": 10,
+    "e": 15,
+    "f": 27,
+    "g": 9,
+    "h": 26,
+    "i": 7,
+    "j": 24,
+    "k": 13,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 12,
+    "p": 16,
+    "q": 29,
+    "r": 21,
+    "s": 18,
+    "t": 8,
+    "u": 17,
+    "v": 28,
+    "w": 22,
+    "x": 32,
+    "y": 19,
+    "z": 30,
+    "|": 5
+  },
+  "che": {
+    "'": 37,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "а": 4,
+    "б": 27,
+    "в": 26,
+    "г": 23,
+    "д": 14,
+    "е": 10,
+    "ж": 33,
+    "з": 28,
+    "и": 12,
+    "й": 21,
+    "к": 18,
+    "л": 7,
+    "м": 24,
+    "н": 8,
+    "о": 15,
+    "п": 31,
+    "р": 13,
+    "с": 22,
+    "т": 17,
+    "у": 9,
+    "ф": 38,
+    "х": 6,
+    "ц": 20,
+    "ч": 25,
+    "ш": 16,
+    "ъ": 29,
+    "ь": 11,
+    "э": 32,
+    "ю": 35,
+    "я": 30,
+    "ӏ": 19,
+    "–": 36
+  },
+  "chf": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 6,
+    "d": 23,
+    "e": 11,
+    "f": 35,
+    "g": 30,
+    "h": 17,
+    "i": 9,
+    "j": 14,
+    "k": 38,
+    "l": 18,
+    "m": 21,
+    "n": 8,
+    "o": 13,
+    "p": 22,
+    "q": 24,
+    "r": 26,
+    "s": 20,
+    "t": 12,
+    "u": 7,
+    "v": 33,
+    "w": 28,
+    "x": 25,
+    "y": 19,
+    "z": 27,
+    "|": 4,
+    "á": 31,
+    "ä": 16,
+    "é": 34,
+    "í": 32,
+    "ñ": 37,
+    "ó": 36,
+    "ú": 29
+  },
+  "chv": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ç": 26,
+    "ă": 22,
+    "ĕ": 20,
+    "а": 5,
+    "б": 38,
+    "в": 21,
+    "г": 37,
+    "д": 35,
+    "е": 7,
+    "ж": 42,
+    "з": 41,
+    "и": 14,
+    "й": 27,
+    "к": 15,
+    "л": 10,
+    "м": 17,
+    "н": 6,
+    "о": 29,
+    "п": 13,
+    "р": 8,
+    "с": 11,
+    "т": 9,
+    "у": 12,
+    "ф": 39,
+    "х": 19,
+    "ц": 40,
+    "ч": 24,
+    "ш": 25,
+    "щ": 44,
+    "ы": 28,
+    "ь": 30,
+    "э": 36,
+    "ю": 34,
+    "я": 31,
+    "ҫ": 23,
+    "ӑ": 16,
+    "ӗ": 18,
+    "ӱ": 46,
+    "ӳ": 33,
+    "–": 45,
+    "—": 43
+  },
+  "chz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 32,
+    "d": 26,
+    "e": 9,
+    "f": 48,
+    "g": 25,
+    "h": 6,
+    "i": 12,
+    "j": 10,
+    "k": 20,
+    "l": 16,
+    "m": 21,
+    "n": 15,
+    "o": 18,
+    "p": 40,
+    "q": 52,
+    "r": 36,
+    "s": 24,
+    "t": 30,
+    "u": 27,
+    "v": 51,
+    "w": 31,
+    "x": 53,
+    "y": 22,
+    "z": 50,
+    "|": 4,
+    "á": 47,
+    "ä": 14,
+    "é": 44,
+    "ë": 23,
+    "í": 45,
+    "ñ": 37,
+    "ó": 46,
+    "ø": 33,
+    "ú": 42,
+    "ɨ": 28,
+    "ʉ": 34,
+    "ʼ": 49,
+    "ˈ": 29,
+    "ˉ": 8,
+    "ˊ": 13,
+    "ˋ": 38,
+    "̃": 39,
+    "̱": 7,
+    "—": 43,
+    "ꜗ": 17,
+    "ꜘ": 19,
+    "ꜙ": 11,
+    "ꜚ": 41
+  },
+  "cjo": {
+    "'": 32,
+    "-": 35,
+    "0": 25,
+    "1": 26,
+    "2": 28,
+    "3": 27,
+    "4": 30,
+    "5": 33,
+    "6": 34,
+    "7": 31,
+    "8": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 23,
+    "e": 8,
+    "g": 21,
+    "h": 17,
+    "i": 5,
+    "j": 19,
+    "k": 11,
+    "m": 15,
+    "n": 10,
+    "o": 12,
+    "p": 13,
+    "r": 9,
+    "s": 16,
+    "t": 7,
+    "u": 38,
+    "v": 29,
+    "w": 18,
+    "y": 14,
+    "z": 20,
+    "|": 6,
+    "ñ": 22,
+    "—": 24
+  },
+  "cjp": {
+    "'": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 27,
+    "d": 24,
+    "e": 8,
+    "f": 38,
+    "g": 36,
+    "h": 23,
+    "i": 6,
+    "j": 12,
+    "k": 7,
+    "l": 16,
+    "m": 21,
+    "n": 17,
+    "o": 22,
+    "p": 26,
+    "q": 41,
+    "r": 15,
+    "s": 11,
+    "t": 13,
+    "u": 19,
+    "v": 37,
+    "w": 14,
+    "x": 43,
+    "y": 20,
+    "z": 39,
+    "|": 4,
+    "­": 45,
+    "à": 30,
+    "á": 29,
+    "ä": 10,
+    "è": 44,
+    "é": 34,
+    "ë": 32,
+    "í": 31,
+    "ñ": 25,
+    "ó": 40,
+    "ö": 28,
+    "ú": 33,
+    "̱": 9,
+    "—": 35
+  },
+  "cjs": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 26,
+    "в": 36,
+    "г": 27,
+    "д": 14,
+    "е": 6,
+    "ж": 32,
+    "з": 25,
+    "и": 7,
+    "й": 24,
+    "к": 20,
+    "л": 12,
+    "м": 22,
+    "н": 9,
+    "о": 16,
+    "п": 13,
+    "р": 8,
+    "с": 15,
+    "т": 11,
+    "у": 23,
+    "ф": 35,
+    "х": 39,
+    "ц": 41,
+    "ч": 18,
+    "ш": 30,
+    "ы": 10,
+    "ь": 40,
+    "э": 33,
+    "я": 37,
+    "ё": 38,
+    "ғ": 19,
+    "қ": 17,
+    "ң": 21,
+    "ӧ": 31,
+    "ӱ": 29,
+    "–": 28
+  },
+  "ckb": {
+    "!": 97,
+    "\"": 69,
+    "%": 96,
+    "'": 77,
+    "+": 103,
+    ",": 76,
+    "-": 52,
+    ".": 46,
+    "/": 64,
+    "0": 38,
+    "1": 41,
+    "2": 45,
+    "3": 55,
+    "4": 54,
+    "5": 49,
+    "6": 53,
+    "7": 65,
+    "8": 61,
+    "9": 58,
+    ":": 68,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 85,
+    "]": 86,
+    "a": 59,
+    "b": 92,
+    "c": 66,
+    "d": 74,
+    "e": 83,
+    "f": 82,
+    "g": 90,
+    "h": 75,
+    "i": 78,
+    "j": 99,
+    "k": 94,
+    "l": 87,
+    "m": 88,
+    "n": 71,
+    "o": 67,
+    "p": 70,
+    "q": 106,
+    "r": 79,
+    "s": 73,
+    "t": 72,
+    "u": 81,
+    "v": 80,
+    "w": 95,
+    "x": 93,
+    "y": 98,
+    "z": 107,
+    "|": 4,
+    "õ": 104,
+    "،": 35,
+    "؛": 89,
+    "؟": 100,
+    "ء": 105,
+    "ؤ": 109,
+    "ئ": 23,
+    "ا": 7,
+    "ب": 14,
+    "ت": 12,
+    "ث": 110,
+    "ج": 30,
+    "ح": 40,
+    "خ": 28,
+    "د": 13,
+    "ر": 10,
+    "ز": 25,
+    "س": 18,
+    "ش": 20,
+    "ع": 62,
+    "غ": 50,
+    "ـ": 102,
+    "ف": 32,
+    "ق": 34,
+    "ك": 33,
+    "ل": 16,
+    "م": 19,
+    "ن": 9,
+    "ه": 22,
+    "و": 8,
+    "ى": 84,
+    "ي": 108,
+    "٠": 37,
+    "١": 39,
+    "٢": 42,
+    "٣": 56,
+    "٤": 51,
+    "٥": 48,
+    "٦": 57,
+    "٧": 63,
+    "٨": 47,
+    "٩": 44,
+    "پ": 24,
+    "چ": 29,
+    "ڕ": 26,
+    "ژ": 31,
+    "ڤ": 36,
+    "ک": 11,
+    "گ": 21,
+    "ڵ": 27,
+    "ھ": 60,
+    "ۆ": 17,
+    "ی": 6,
+    "ێ": 15,
+    "ە": 5,
+    "‌": 101,
+    "‎": 43,
+    "‏": 91
+  },
+  "cko": {
+    "'": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 30,
+    "d": 20,
+    "e": 18,
+    "f": 22,
+    "g": 24,
+    "h": 29,
+    "i": 6,
+    "j": 26,
+    "k": 12,
+    "l": 27,
+    "m": 8,
+    "n": 7,
+    "o": 14,
+    "p": 28,
+    "r": 13,
+    "s": 16,
+    "t": 19,
+    "u": 10,
+    "v": 32,
+    "w": 21,
+    "y": 17,
+    "z": 31,
+    "|": 4,
+    "á": 25,
+    "é": 37,
+    "í": 34,
+    "ó": 33,
+    "ú": 35,
+    "ŋ": 23,
+    "ɔ": 15,
+    "ɛ": 9,
+    "́": 36
+  },
+  "ckt": {
+    "'": 34,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "а": 9,
+    "б": 39,
+    "в": 16,
+    "г": 15,
+    "д": 35,
+    "е": 24,
+    "ж": 40,
+    "з": 37,
+    "и": 10,
+    "й": 25,
+    "к": 11,
+    "л": 33,
+    "м": 14,
+    "н": 6,
+    "о": 20,
+    "п": 26,
+    "р": 12,
+    "с": 28,
+    "т": 8,
+    "у": 21,
+    "ф": 36,
+    "х": 38,
+    "ц": 41,
+    "ч": 23,
+    "ъ": 17,
+    "ы": 4,
+    "ь": 22,
+    "э": 7,
+    "ю": 32,
+    "я": 29,
+    "ё": 27,
+    "ӄ": 19,
+    "ӈ": 18,
+    "ԓ": 13,
+    "–": 31
+  },
+  "cla": {
+    "'": 25,
+    "-": 32,
+    "0": 34,
+    "2": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 28,
+    "d": 23,
+    "e": 12,
+    "f": 19,
+    "g": 20,
+    "h": 14,
+    "i": 6,
+    "j": 29,
+    "k": 13,
+    "l": 16,
+    "m": 7,
+    "n": 9,
+    "o": 18,
+    "p": 27,
+    "r": 21,
+    "s": 8,
+    "t": 10,
+    "u": 11,
+    "v": 31,
+    "w": 15,
+    "y": 17,
+    "z": 30,
+    "|": 4,
+    "ɓ": 22,
+    "ɗ": 24
+  },
+  "cle": {
+    "'": 46,
+    "-": 45,
+    "1": 25,
+    "2": 10,
+    "3": 5,
+    "4": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 30,
+    "c": 17,
+    "d": 15,
+    "e": 18,
+    "f": 34,
+    "g": 26,
+    "h": 6,
+    "i": 8,
+    "j": 14,
+    "k": 44,
+    "l": 20,
+    "m": 23,
+    "n": 19,
+    "o": 27,
+    "p": 39,
+    "q": 43,
+    "r": 37,
+    "s": 21,
+    "t": 32,
+    "u": 22,
+    "v": 38,
+    "x": 24,
+    "y": 16,
+    "z": 42,
+    "|": 4,
+    "á": 11,
+    "é": 29,
+    "í": 28,
+    "ñ": 35,
+    "ó": 36,
+    "ú": 33,
+    "ü": 41,
+    "ŋ": 40,
+    "ɨ": 13,
+    "́": 31,
+    "̱": 12
+  },
+  "cly": {
+    "'": 7,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 26,
+    "c": 10,
+    "d": 18,
+    "e": 13,
+    "f": 33,
+    "g": 20,
+    "h": 17,
+    "i": 9,
+    "j": 22,
+    "k": 38,
+    "l": 15,
+    "m": 24,
+    "n": 5,
+    "o": 12,
+    "p": 27,
+    "q": 25,
+    "r": 19,
+    "s": 14,
+    "t": 11,
+    "u": 8,
+    "v": 34,
+    "x": 23,
+    "y": 16,
+    "z": 37,
+    "|": 4,
+    "á": 36,
+    "é": 32,
+    "í": 30,
+    "ñ": 21,
+    "ó": 35,
+    "ú": 28,
+    "ü": 31,
+    "́": 39,
+    "—": 29
+  },
+  "cme": {
+    "'": 36,
+    "-": 16,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 28,
+    "d": 17,
+    "e": 12,
+    "f": 30,
+    "g": 24,
+    "h": 27,
+    "i": 6,
+    "j": 33,
+    "k": 22,
+    "l": 14,
+    "m": 10,
+    "n": 9,
+    "o": 13,
+    "p": 32,
+    "r": 20,
+    "s": 21,
+    "t": 25,
+    "u": 7,
+    "v": 37,
+    "w": 29,
+    "y": 19,
+    "|": 4,
+    "ã": 26,
+    "ĩ": 31,
+    "ŋ": 8,
+    "ũ": 34,
+    "ɔ": 15,
+    "ɛ": 18,
+    "̃": 35,
+    "‐": 23,
+    "—": 38
+  },
+  "cmn-script_simplified": {
+    "%": 899,
+    "&": 2065,
+    "'": 1415,
+    "+": 1936,
+    ",": 479,
+    "-": 684,
+    ".": 916,
+    "/": 1405,
+    "0": 14,
+    "1": 21,
+    "2": 49,
+    "3": 132,
+    "4": 131,
+    "5": 98,
+    "6": 150,
+    "7": 211,
+    "8": 149,
+    "9": 86,
+    ":": 1176,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 13,
+    "b": 235,
+    "c": 105,
+    "d": 128,
+    "e": 27,
+    "f": 308,
+    "g": 119,
+    "h": 88,
+    "i": 48,
+    "j": 491,
+    "k": 212,
+    "l": 59,
+    "m": 133,
+    "n": 34,
+    "o": 35,
+    "p": 192,
+    "q": 1508,
+    "r": 36,
+    "s": 42,
+    "t": 57,
+    "u": 116,
+    "v": 402,
+    "w": 423,
+    "x": 811,
+    "y": 309,
+    "z": 757,
+    "°": 1648,
+    "·": 1691,
+    "é": 1911,
+    "í": 1434,
+    "ó": 2046,
+    "õ": 1632,
+    "ł": 2310,
+    "̃": 3598,
+    "̌": 3506,
+    "μ": 3818,
+    "и": 3597,
+    "т": 3137,
+    "—": 1616,
+    "”": 1078,
+    "•": 1475,
+    "、": 2125,
+    "《": 2294,
+    "》": 2295,
+    "オ": 4215,
+    "カ": 4216,
+    "ド": 4218,
+    "ヤ": 4217,
+    "ㄟ": 3869,
+    "䴕": 4490,
+    "一": 6,
+    "丁": 939,
+    "七": 983,
+    "万": 378,
+    "丈": 1865,
+    "三": 294,
+    "上": 29,
+    "下": 120,
+    "不": 23,
+    "与": 109,
+    "丑": 1934,
+    "专": 671,
+    "且": 386,
+    "丕": 3315,
+    "世": 108,
+    "丘": 1326,
+    "丙": 2500,
+    "业": 270,
+    "丛": 1660,
+    "东": 440,
+    "丝": 1163,
+    "丞": 2586,
+    "丟": 3523,
+    "丢": 1484,
+    "两": 312,
+    "严": 525,
+    "丧": 1417,
+    "个": 31,
+    "丫": 3061,
+    "中": 30,
+    "丰": 1301,
+    "串": 2048,
+    "临": 819,
+    "丸": 2635,
+    "丹": 807,
+    "为": 15,
+    "主": 184,
+    "丽": 1345,
+    "举": 320,
+    "乂": 2863,
+    "乃": 2279,
+    "久": 697,
+    "么": 371,
+    "义": 551,
+    "之": 65,
+    "乌": 1373,
+    "乍": 2318,
+    "乎": 618,
+    "乏": 1338,
+    "乐": 461,
+    "乒": 3222,
+    "乓": 3429,
+    "乔": 941,
+    "乖": 4050,
+    "乘": 663,
+    "乙": 2389,
+    "九": 1007,
+    "乞": 2017,
+    "也": 66,
+    "习": 446,
+    "乡": 1230,
+    "书": 556,
+    "买": 799,
+    "乱": 1116,
+    "乳": 1493,
+    "乾": 2401,
+    "了": 10,
+    "予": 1676,
+    "争": 488,
+    "事": 190,
+    "二": 265,
+    "于": 37,
+    "亏": 1974,
+    "云": 1179,
+    "互": 554,
+    "五": 833,
+    "井": 2382,
+    "亚": 67,
+    "些": 54,
+    "亡": 726,
+    "交": 305,
+    "亥": 2047,
+    "亦": 1429,
+    "产": 266,
+    "亨": 1598,
+    "亩": 3347,
+    "享": 921,
+    "京": 934,
+    "亭": 2104,
+    "亮": 1629,
+    "亲": 900,
+    "亳": 3760,
+    "人": 11,
+    "亿": 839,
+    "什": 555,
+    "仁": 1556,
+    "仄": 3964,
+    "仅": 531,
+    "仆": 2641,
+    "仇": 1695,
+    "今": 337,
+    "介": 1949,
+    "仍": 627,
+    "从": 136,
+    "仑": 1806,
+    "仓": 2405,
+    "仔": 1794,
+    "仕": 2438,
+    "他": 20,
+    "仗": 3328,
+    "付": 999,
+    "仙": 2393,
+    "仞": 4089,
+    "代": 260,
+    "���": 628,
+    "以": 22,
+    "仪": 1199,
+    "们": 33,
+    "仰": 1426,
+    "仲": 2494,
+    "件": 243,
+    "价": 472,
+    "任": 277,
+    "份": 453,
+    "仿": 1722,
+    "企": 1515,
+    "伊": 362,
+    "伍": 2197,
+    "伎": 3807,
+    "伏": 1569,
+    "伐": 1649,
+    "休": 1198,
+    "众": 593,
+    "优": 856,
+    "伙": 2922,
+    "会": 25,
+    "伛": 4211,
+    "伞": 2492,
+    "伟": 1977,
+    "传": 353,
+    "伤": 516,
+    "伦": 608,
+    "伪": 2750,
+    "伫": 4460,
+    "伯": 654,
+    "估": 818,
+    "伴": 1680,
+    "伸": 1463,
+    "伺": 3199,
+    "似": 732,
+    "伽": 1451,
+    "佃": 3570,
+    "但": 80,
+    "位": 188,
+    "低": 553,
+    "住": 317,
+    "佐": 1509,
+    "佑": 2526,
+    "体": 126,
+    "何": 383,
+    "佘": 4173,
+    "余": 901,
+    "佚": 3851,
+    "佛": 1003,
+    "作": 101,
+    "佟": 3730,
+    "你": 92,
+    "佣": 2089,
+    "佤": 4051,
+    "佥": 3027,
+    "佩": 1697,
+    "佬": 4018,
+    "佯": 4434,
+    "佰": 3268,
+    "佳": 898,
+    "佼": 4054,
+    "使": 147,
+    "侂": 4132,
+    "侃": 3107,
+    "侄": 2602,
+    "侈": 2254,
+    "例": 510,
+    "侍": 2436,
+    "侏": 3547,
+    "侗": 3491,
+    "供": 283,
+    "依": 915,
+    "侠": 2835,
+    "侣": 3083,
+    "侦": 2247,
+    "侧": 1140,
+    "侨": 2684,
+    "侪": 4125,
+    "侬": 3583,
+    "侮": 3991,
+    "侯": 1983,
+    "侵": 769,
+    "便": 548,
+    "促": 1107,
+    "俄": 725,
+    "俅": 4312,
+    "俊": 2558,
+    "俐": 4122,
+    "俗": 950,
+    "俘": 2259,
+    "俚": 1875,
+    "保": 218,
+    "俞": 2978,
+    "信": 380,
+    "俣": 3952,
+    "俨": 3373,
+    "俩": 3322,
+    "俭": 3024,
+    "修": 643,
+    "俯": 2042,
+    "俱": 1067,
+    "俳": 3852,
+    "俵": 4094,
+    "俸": 3866,
+    "倍": 1192,
+    "倒": 1769,
+    "倘": 3974,
+    "候": 502,
+    "倚": 4262,
+    "借": 1355,
+    "倡": 2718,
+    "倦": 4388,
+    "倩": 4282,
+    "倪": 3157,
+    "倭": 3365,
+    "债": 2153,
+    "倻": 4276,
+    "值": 925,
+    "倾": 1222,
+    "偃": 4170,
+    "假": 1375,
+    "偈": 4351,
+    "偏": 1607,
+    "偕": 3893,
+    "做": 322,
+    "停": 812,
+    "健": 1038,
+    "偲": 3931,
+    "偶": 1781,
+    "偷": 1542,
+    "偿": 3152,
+    "傀": 3496,
+    "傅": 2499,
+    "傉": 4321,
+    "傍": 3850,
+    "傕": 4037,
+    "傣": 3174,
+    "储": 1409,
+    "傩": 4066,
+    "催": 2584,
+    "傲": 3012,
+    "像": 382,
+    "僖": 3590,
+    "僚": 3176,
+    "僧": 1862,
+    "僵": 3441,
+    "儆": 4032,
+    "儋": 4021,
+    "儒": 2590,
+    "儡": 3497,
+    "儿": 810,
+    "兀": 3254,
+    "允": 1168,
+    "元": 397,
+    "兄": 1857,
+    "充": 1053,
+    "兆": 2575,
+    "先": 341,
+    "光": 331,
+    "克": 115,
+    "免": 782,
+    "兑": 1810,
+    "兔": 2107,
+    "兖": 2958,
+    "党": 979,
+    "兜": 3080,
+    "入": 159,
+    "全": 198,
+    "八": 1300,
+    "公": 63,
+    "六": 928,
+    "兮": 3515,
+    "兰": 214,
+    "共": 407,
+    "关": 222,
+    "兴": 1070,
+    "兵": 974,
+    "其": 45,
+    "具": 493,
+    "典": 762,
+    "兹": 1606,
+    "养": 743,
+    "兼": 1730,
+    "兽": 2477,
+    "冀": 3076,
+    "内": 186,
+    "冈": 1393,
+    "冉": 3788,
+    "册": 1449,
+    "再": 669,
+    "冒": 1962,
+    "冕": 2512,
+    "写": 660,
+    "军": 245,
+    "农": 932,
+    "冠": 944,
+    "冢": 2865,
+    "冤": 4268,
+    "冥": 3686,
+    "冬": 958,
+    "冯": 2339,
+    "冰": 418,
+    "冲": 967,
+    "决": 350,
+    "况": 336,
+    "冶": 2717,
+    "冷": 904,
+    "冻": 2265,
+    "冼": 4224,
+    "净": 1332,
+    "凄": 3280,
+    "准": 478,
+    "凉": 1622,
+    "凋": 3172,
+    "凌": 1641,
+    "减": 1074,
+    "凑": 3191,
+    "凝": 2704,
+    "几": 290,
+    "凡": 2324,
+    "凤": 2366,
+    "凭": 2220,
+    "凯": 1397,
+    "凰": 2538,
+    "凳": 3673,
+    "凶": 2264,
+    "凸": 2770,
+    "凹": 1906,
+    "出": 56,
+    "击": 340,
+    "函": 1873,
+    "凿": 3048,
+    "刀": 989,
+    "刃": 1143,
+    "分": 75,
+    "切": 583,
+    "刈": 3609,
+    "刊": 1637,
+    "刍": 4183,
+    "刑": 1423,
+    "划": 519,
+    "列": 585,
+    "刘": 2358,
+    "则": 521,
+    "刚": 1104,
+    "创": 503,
+    "初": 588,
+    "删": 1844,
+    "判": 808,
+    "別": 2649,
+    "利": 78,
+    "别": 390,
+    "刮": 3902,
+    "到": 28,
+    "制": 240,
+    "刷": 1723,
+    "券": 2154,
+    "刹": 3159,
+    "刺": 1271,
+    "刻": 1325,
+    "剀": 4381,
+    "剂": 1557,
+    "剃": 4365,
+    "削": 1356,
+    "剌": 3270,
+    "前": 73,
+    "剑": 1476,
+    "剖": 2007,
+    "剥": 1458,
+    "剧": 1054,
+    "剩": 1282,
+    "剪": 1410,
+    "副": 1052,
+    "割": 2252,
+    "剿": 2957,
+    "劈": 3101,
+    "力": 162,
+    "劝": 2821,
+    "办": 602,
+    "功": 667,
+    "加": 110,
+    "务": 310,
+    "劣": 1299,
+    "动": 52,
+    "助": 404,
+    "努": 1157,
+    "劫": 1890,
+    "励": 1535,
+    "劲": 2956,
+    "劳": 1523,
+    "劾": 3039,
+    "势": 773,
+    "勃": 2097,
+    "勇": 2142,
+    "勉": 3179,
+    "勋": 1864,
+    "勍": 3839,
+    "勐": 3195,
+    "勒": 730,
+    "勘": 1675,
+    "募": 2031,
+    "勤": 1784,
+    "勺": 4133,
+    "勾": 1933,
+    "勿": 1990,
+    "匀": 3458,
+    "包": 373,
+    "匆": 3676,
+    "匈": 1498,
+    "匍": 3463,
+    "匐": 3269,
+    "匕": 3593,
+    "化": 185,
+    "北": 288,
+    "匙": 2321,
+    "匝": 2994,
+    "匠": 2845,
+    "匡": 2965,
+    "匣": 2187,
+    "匪": 3100,
+    "匮": 4413,
+    "匹": 1521,
+    "区": 117,
+    "医": 477,
+    "匾": 4442,
+    "匿": 3146,
+    "十": 302,
+    "千": 610,
+    "升": 917,
+    "午": 906,
+    "卉": 1817,
+    "半": 814,
+    "华": 537,
+    "协": 698,
+    "卑": 2521,
+    "卒": 2431,
+    "卓": 2109,
+    "单": 505,
+    "卖": 2049,
+    "南": 291,
+    "博": 528,
+    "卜": 2593,
+    "卞": 3521,
+    "占": 1088,
+    "卡": 354,
+    "卢": 857,
+    "卤": 3811,
+    "卦": 3290,
+    "卧": 1931,
+    "卫": 536,
+    "卯": 2566,
+    "印": 316,
+    "危": 701,
+    "即": 443,
+    "却": 718,
+    "卵": 1743,
+    "卷": 1083,
+    "卸": 1679,
+    "卿": 2464,
+    "厂": 1626,
+    "厄": 1344,
+    "厅": 913,
+    "历": 376,
+    "厉": 1540,
+    "压": 894,
+    "厌": 1771,
+    "厍": 3474,
+    "厕": 1978,
+    "厘": 1998,
+    "厚": 1465,
+    "厝": 2823,
+    "原": 204,
+    "厢": 1323,
+    "厥": 2709,
+    "厦": 2445,
+    "厨": 2950,
+    "厩": 4155,
+    "去": 244,
+    "县": 1999,
+    "参": 364,
+    "叅": 4178,
+    "又": 795,
+    "叉": 2483,
+    "及": 289,
+    "友": 907,
+    "双": 825,
+    "反": 466,
+    "发": 38,
+    "叔": 2620,
+    "取": 346,
+    "受": 154,
+    "变": 239,
+    "叙": 2394,
+    "叛": 2561,
+    "叟": 3887,
+    "叠": 1341,
+    "口": 366,
+    "古": 417,
+    "句": 1071,
+    "另": 737,
+    "叩": 3833,
+    "只": 221,
+    "叫": 817,
+    "召": 2124,
+    "叭": 3565,
+    "可": 19,
+    "台": 441,
+    "史": 410,
+    "右": 735,
+    "叶": 1262,
+    "号": 389,
+    "司": 201,
+    "叹": 2013,
+    "叻": 3295,
+    "吁": 1950,
+    "吃": 785,
+    "各": 391,
+    "合": 272,
+    "吉": 879,
+    "吊": 1659,
+    "同": 127,
+    "名": 122,
+    "后": 62,
+    "吏": 2683,
+    "吐": 1524,
+    "向": 238,
+    "吓": 1913,
+    "吕": 2388,
+    "吗": 1564,
+    "君": 1235,
+    "吞": 1699,
+    "吟": 3548,
+    "吠": 3715,
+    "否": 897,
+    "吧": 1414,
+    "吨": 3438,
+    "含": 1085,
+    "听": 982,
+    "启": 1111,
+    "吴": 2000,
+    "吵": 2119,
+    "吸": 993,
+    "吹": 2335,
+    "吻": 1996,
+    "吼": 2099,
+    "吾": 2008,
+    "呀": 3224,
+    "呆": 3543,
+    "呈": 1130,
+    "告": 429,
+    "呋": 4152,
+    "呐": 4057,
+    "呔": 4025,
+    "呕": 2037,
+    "员": 195,
+    "呢": 2186,
+    "呤": 3536,
+    "周": 285,
+    "味": 863,
+    "呵": 4165,
+    "呻": 4238,
+    "呼": 1094,
+    "命": 565,
+    "咀": 3629,
+    "咆": 4005,
+    "和": 9,
+    "咎": 1681,
+    "咏": 3021,
+    "咒": 2085,
+    "咕": 3058,
+    "咖": 1274,
+    "咙": 1943,
+    "咨": 1804,
+    "咪": 4101,
+    "咬": 1614,
+    "咸": 2458,
+    "咽": 3473,
+    "哀": 3170,
+    "品": 482,
+    "哄": 3911,
+    "哇": 2234,
+    "哈": 694,
+    "哉": 3189,
+    "响": 384,
+    "哑": 4223,
+    "哔": 3750,
+    "哥": 612,
+    "哦": 2141,
+    "哨": 3589,
+    "哩": 2895,
+    "哪": 1358,
+    "哭": 2225,
+    "哮": 3626,
+    "哲": 2245,
+    "哺": 1845,
+    "哼": 4114,
+    "唁": 4306,
+    "唆": 4139,
+    "唇": 2408,
+    "唐": 1938,
+    "唑": 3308,
+    "唛": 3889,
+    "唤": 3198,
+    "售": 1101,
+    "唯": 903,
+    "唱": 1346,
+    "唾": 3865,
+    "啄": 2949,
+    "商": 512,
+    "啉": 4091,
+    "啊": 2692,
+    "啡": 1329,
+    "啤": 2112,
+    "啦": 2979,
+    "啧": 3674,
+    "啮": 1823,
+    "啰": 3507,
+    "啶": 3251,
+    "啸": 1170,
+    "喀": 2071,
+    "喂": 2175,
+    "喃": 4153,
+    "善": 1343,
+    "喇": 2714,
+    "喉": 1942,
+    "喊": 3618,
+    "喔": 3194,
+    "喘": 2146,
+    "喙": 2081,
+    "喜": 987,
+    "喝": 2696,
+    "喧": 4177,
+    "喱": 1980,
+    "喵": 3671,
+    "喷": 1286,
+    "喹": 4090,
+    "喻": 2982,
+    "喽": 3894,
+    "嗓": 3264,
+    "嗜": 2890,
+    "嗣": 2608,
+    "嗲": 3814,
+    "嘉": 2352,
+    "嘌": 3535,
+    "嘎": 3248,
+    "嘛": 3065,
+    "嘟": 3967,
+    "嘧": 3250,
+    "嘲": 1663,
+    "嘴": 1801,
+    "嘻": 4266,
+    "噌": 4470,
+    "噜": 2880,
+    "器": 524,
+    "噩": 3489,
+    "噪": 3643,
+    "噬": 1715,
+    "噱": 2331,
+    "噶": 2651,
+    "嚓": 4141,
+    "嚣": 4454,
+    "囊": 1863,
+    "囚": 1693,
+    "四": 490,
+    "回": 325,
+    "因": 90,
+    "团": 522,
+    "园": 498,
+    "困": 1029,
+    "围": 533,
+    "固": 1256,
+    "国": 18,
+    "图": 323,
+    "圃": 4476,
+    "圆": 2363,
+    "圈": 1740,
+    "圜": 3981,
+    "土": 377,
+    "圣": 456,
+    "在": 5,
+    "圩": 3922,
+    "圪": 3944,
+    "圭": 1543,
+    "地": 17,
+    "圳": 2822,
+    "圹": 4083,
+    "场": 123,
+    "圻": 3384,
+    "圾": 2251,
+    "址": 1062,
+    "坂": 2647,
+    "均": 1263,
+    "坊": 2583,
+    "坍": 1708,
+    "坎": 1519,
+    "坏": 642,
+    "坐": 804,
+    "坑": 1692,
+    "块": 849,
+    "坚": 1097,
+    "坛": 1709,
+    "坜": 3994,
+    "坝": 2625,
+    "坞": 2617,
+    "坟": 1579,
+    "坠": 1087,
+    "坡": 952,
+    "坤": 2929,
+    "坦": 575,
+    "坨": 4079,
+    "坪": 2642,
+    "坯": 3989,
+    "坳": 3670,
+    "坷": 2236,
+    "垂": 2488,
+    "垃": 2250,
+    "垄": 3098,
+    "型": 313,
+    "垒": 1295,
+    "垠": 2095,
+    "垢": 3905,
+    "垣": 2768,
+    "垦": 2753,
+    "垩": 2253,
+    "垫": 1501,
+    "埃": 639,
+    "埇": 4408,
+    "埈": 4072,
+    "埋": 2702,
+    "城": 248,
+    "埔": 1371,
+    "埕": 4060,
+    "埗": 4326,
+    "域": 587,
+    "埠": 2715,
+    "培": 1439,
+    "基": 246,
+    "堂": 1298,
+    "堆": 1478,
+    "堇": 2446,
+    "堕": 1485,
+    "堡": 425,
+    "堤": 1951,
+    "堪": 2132,
+    "堰": 3787,
+    "堵": 1445,
+    "塌": 1628,
+    "塑": 1609,
+    "塔": 455,
+    "塘": 2455,
+    "塞": 517,
+    "填": 1257,
+    "塬": 4384,
+    "塾": 3853,
+    "墀": 4315,
+    "境": 676,
+    "墅": 2757,
+    "墓": 1014,
+    "墙": 1215,
+    "增": 734,
+    "墟": 2326,
+    "墨": 1337,
+    "墩": 3134,
+    "壁": 1217,
+    "壤": 1820,
+    "士": 351,
+    "壬": 2587,
+    "壮": 1762,
+    "声": 549,
+    "壳": 1621,
+    "壶": 2820,
+    "壸": 3560,
+    "壹": 3660,
+    "壽": 4464,
+    "处": 276,
+    "备": 560,
+    "复": 541,
+    "夏": 668,
+    "夕": 1514,
+    "外": 207,
+    "多": 39,
+    "夜": 843,
+    "够": 670,
+    "夥": 3664,
+    "大": 16,
+    "天": 158,
+    "太": 328,
+    "夫": 457,
+    "夭": 3519,
+    "央": 1324,
+    "夯": 3782,
+    "失": 468,
+    "头": 363,
+    "夷": 1466,
+    "夸": 1793,
+    "夹": 1848,
+    "夺": 1080,
+    "奂": 4419,
+    "奇": 733,
+    "奈": 2369,
+    "奉": 1880,
+    "奋": 1883,
+    "奎": 1571,
+    "奏": 1288,
+    "契": 2453,
+    "奔": 1682,
+    "奕": 3698,
+    "奖": 954,
+    "套": 2278,
+    "奘": 3516,
+    "奚": 3858,
+    "奠": 3151,
+    "奢": 1826,
+    "奣": 4194,
+    "奥": 259,
+    "女": 607,
+    "奴": 2427,
+    "奶": 1560,
+    "奸": 2179,
+    "她": 598,
+    "好": 311,
+    "如": 96,
+    "妃": 2563,
+    "妆": 2886,
+    "妇": 1073,
+    "妈": 2515,
+    "妊": 3581,
+    "妍": 4414,
+    "妒": 4416,
+    "妓": 3267,
+    "妖": 2699,
+    "妙": 2343,
+    "妡": 4174,
+    "妤": 3726,
+    "妥": 2937,
+    "妨": 1690,
+    "妮": 1494,
+    "妲": 3968,
+    "妳": 2653,
+    "妹": 1782,
+    "妻": 1244,
+    "妾": 2909,
+    "姆": 657,
+    "姊": 2345,
+    "始": 319,
+    "姐": 1889,
+    "姑": 2674,
+    "姒": 3758,
+    "姓": 1732,
+    "委": 850,
+    "姚": 2528,
+    "姜": 2432,
+    "姨": 4000,
+    "姬": 2291,
+    "姮": 4398,
+    "姻": 1454,
+    "姿": 1096,
+    "威": 420,
+    "娃": 2564,
+    "娄": 3031,
+    "娅": 1744,
+    "娆": 3990,
+    "娇": 4212,
+    "娘": 2508,
+    "娜": 1381,
+    "娟": 3910,
+    "娠": 3582,
+    "娣": 3962,
+    "娥": 4087,
+    "娱": 1526,
+    "娶": 2686,
+    "婆": 2003,
+    "婉": 3747,
+    "婕": 3725,
+    "婚": 1136,
+    "婢": 4462,
+    "婪": 2134,
+    "婴": 2270,
+    "婵": 3734,
+    "婷": 3691,
+    "婺": 4274,
+    "婿": 3002,
+    "媒": 997,
+    "媚": 2271,
+    "媛": 3105,
+    "媲": 1620,
+    "媳": 3366,
+    "嫁": 2588,
+    "嫉": 4244,
+    "嫌": 1177,
+    "嫔": 3212,
+    "嫖": 4197,
+    "嫡": 2984,
+    "嫣": 3932,
+    "嫩": 2341,
+    "嬤": 4311,
+    "嬴": 4393,
+    "子": 87,
+    "孔": 2148,
+    "孕": 2923,
+    "孖": 4063,
+    "字": 408,
+    "存": 345,
+    "孙": 1650,
+    "孚": 2838,
+    "孛": 4489,
+    "孜": 3169,
+    "孝": 2450,
+    "孟": 2289,
+    "孢": 2511,
+    "季": 374,
+    "孤": 2767,
+    "学": 71,
+    "孩": 509,
+    "孪": 3273,
+    "孵": 3276,
+    "孺": 3568,
+    "宁": 1390,
+    "它": 107,
+    "宅": 1773,
+    "宇": 1012,
+    "守": 1249,
+    "安": 333,
+    "宋": 2356,
+    "完": 495,
+    "宏": 2272,
+    "宕": 3283,
+    "宗": 673,
+    "官": 428,
+    "宙": 1507,
+    "定": 141,
+    "宛": 1935,
+    "宜": 1246,
+    "宝": 1185,
+    "实": 281,
+    "宠": 1839,
+    "审": 940,
+    "客": 370,
+    "宣": 706,
+    "室": 998,
+    "宥": 4395,
+    "宦": 3379,
+    "宪": 1149,
+    "宫": 2325,
+    "宰": 1613,
+    "害": 740,
+    "宴": 1482,
+    "宵": 4286,
+    "家": 50,
+    "容": 480,
+    "宽": 1187,
+    "宾": 1412,
+    "宿": 832,
+    "寀": 4192,
+    "寂": 1869,
+    "寄": 1131,
+    "寅": 2784,
+    "密": 680,
+    "寇": 3156,
+    "富": 496,
+    "寒": 1351,
+    "寓": 1387,
+    "寝": 4284,
+    "寞": 3350,
+    "察": 539,
+    "寡": 3563,
+    "寥": 3744,
+    "寨": 1314,
+    "��": 3036,
+    "寰": 2164,
+    "寸": 1617,
+    "对": 60,
+    "寺": 959,
+    "寻": 896,
+    "导": 379,
+    "寿": 2418,
+    "封": 1015,
+    "専": 4111,
+    "射": 613,
+    "将": 151,
+    "尉": 2496,
+    "尊": 1401,
+    "小": 124,
+    "少": 301,
+    "尔": 94,
+    "尖": 2224,
+    "尘": 1706,
+    "尚": 945,
+    "尝": 1120,
+    "尤": 756,
+    "尧": 2834,
+    "尨": 4295,
+    "就": 93,
+    "尸": 1785,
+    "尹": 2576,
+    "尺": 1312,
+    "尻": 4299,
+    "尼": 205,
+    "尽": 307,
+    "尾": 1802,
+    "尿": 1212,
+    "局": 442,
+    "层": 889,
+    "居": 421,
+    "屈": 2226,
+    "屉": 3987,
+    "届": 1745,
+    "屋": 1025,
+    "屏": 1912,
+    "屐": 4440,
+    "屑": 3380,
+    "展": 542,
+    "属": 712,
+    "屠": 1612,
+    "屡": 2574,
+    "履": 3020,
+    "屯": 2616,
+    "山": 297,
+    "屹": 4142,
+    "屿": 964,
+    "岁": 771,
+    "岈": 4343,
+    "岐": 2748,
+    "岑": 2626,
+    "岔": 4002,
+    "岖": 2040,
+    "岗": 2533,
+    "岚": 3748,
+    "岛": 228,
+    "岩": 927,
+    "岫": 4270,
+    "岬": 3041,
+    "岭": 2039,
+    "岱": 3028,
+    "岳": 2457,
+    "岷": 3104,
+    "岸": 984,
+    "峄": 3723,
+    "峒": 4074,
+    "峙": 3475,
+    "峡": 1110,
+    "峥": 4058,
+    "峨": 2733,
+    "峪": 4233,
+    "峭": 1216,
+    "峯": 4134,
+    "峰": 1254,
+    "峻": 3353,
+    "崁": 4252,
+    "崇": 1562,
+    "崎": 1491,
+    "崔": 2185,
+    "崖": 2044,
+    "崛": 3636,
+    "崧": 4016,
+    "崩": 1727,
+    "崴": 3235,
+    "嵋": 4222,
+    "嵌": 2989,
+    "嵖": 4342,
+    "嵗": 4195,
+    "嵩": 2829,
+    "嵯": 3638,
+    "嵴": 4366,
+    "嶷": 4406,
+    "巅": 2237,
+    "川": 1450,
+    "州": 314,
+    "巡": 1627,
+    "巢": 1477,
+    "工": 269,
+    "左": 765,
+    "巧": 1583,
+    "巨": 1000,
+    "巩": 2338,
+    "巫": 2562,
+    "差": 1023,
+    "巯": 3929,
+    "己": 334,
+    "已": 231,
+    "巳": 3615,
+    "巴": 261,
+    "巷": 1997,
+    "巽": 3804,
+    "巾": 3656,
+    "币": 1623,
+    "市": 268,
+    "布": 199,
+    "帅": 2025,
+    "帆": 2726,
+    "师": 770,
+    "希": 459,
+    "帐": 1599,
+    "帕": 682,
+    "帖": 4191,
+    "帘": 3088,
+    "帚": 4059,
+    "帛": 3229,
+    "帜": 2023,
+    "帝": 753,
+    "带": 330,
+    "帧": 3344,
+    "席": 1109,
+    "帮": 545,
+    "帷": 4028,
+    "常": 72,
+    "帽": 2661,
+    "幂": 3532,
+    "幅": 1441,
+    "幌": 3093,
+    "幔": 4029,
+    "幕": 1002,
+    "幡": 3291,
+    "幢": 2962,
+    "干": 1236,
+    "平": 349,
+    "年": 41,
+    "并": 100,
+    "幸": 1119,
+    "幺": 4322,
+    "幻": 2597,
+    "幼": 2400,
+    "幽": 1767,
+    "广": 591,
+    "庄": 1878,
+    "庆": 1132,
+    "庇": 2881,
+    "床": 1034,
+    "序": 717,
+    "庐": 2825,
+    "库": 1138,
+    "应": 203,
+    "底": 636,
+    "店": 655,
+    "庙": 703,
+    "庚": 2645,
+    "府": 414,
+    "庞": 1724,
+    "废": 1926,
+    "度": 121,
+    "座": 388,
+    "庭": 589,
+    "庵": 2968,
+    "庶": 2831,
+    "康": 1197,
+    "庸": 2836,
+    "庹": 3727,
+    "庾": 4033,
+    "廆": 3955,
+    "廉": 1394,
+    "廊": 2518,
+    "廓": 3130,
+    "廖": 2833,
+    "廪": 3602,
+    "延": 1055,
+    "廷": 1280,
+    "建": 249,
+    "廿": 2940,
+    "开": 125,
+    "弁": 3983,
+    "异": 930,
+    "弃": 1590,
+    "弄": 1971,
+    "弈": 3375,
+    "弊": 3351,
+    "式": 181,
+    "弓": 1098,
+    "引": 426,
+    "弗": 666,
+    "弘": 2414,
+    "弟": 1858,
+    "张": 638,
+    "弢": 3672,
+    "弥": 1929,
+    "弦": 2072,
+    "弧": 2850,
+    "弯": 1174,
+    "弱": 1069,
+    "弹": 1165,
+    "强": 405,
+    "弼": 2737,
+    "归": 1050,
+    "当": 112,
+    "录": 646,
+    "彗": 2038,
+    "彝": 2796,
+    "形": 342,
+    "彤": 4480,
+    "彦": 2491,
+    "彧": 3518,
+    "彩": 1004,
+    "彪": 4078,
+    "彬": 2926,
+    "彭": 2456,
+    "彰": 1891,
+    "影": 315,
+    "役": 1432,
+    "彻": 1124,
+    "彼": 1121,
+    "往": 241,
+    "征": 1044,
+    "径": 1372,
+    "待": 779,
+    "很": 106,
+    "徊": 4290,
+    "律": 436,
+    "徐": 2387,
+    "徒": 741,
+    "得": 82,
+    "徘": 4289,
+    "徙": 1896,
+    "御": 1655,
+    "循": 1661,
+    "徬": 3840,
+    "微": 700,
+    "德": 160,
+    "徽": 2375,
+    "心": 224,
+    "必": 332,
+    "忆": 1725,
+    "忌": 2803,
+    "忍": 2536,
+    "忒": 1928,
+    "志": 648,
+    "忘": 1750,
+    "忙": 1534,
+    "忠": 2433,
+    "忤": 4104,
+    "忧": 1315,
+    "快": 558,
+    "忱": 3873,
+    "念": 727,
+    "忻": 3720,
+    "忽": 2195,
+    "怀": 1348,
+    "态": 584,
+    "怎": 1420,
+    "怒": 2719,
+    "怕": 1431,
+    "怖": 1749,
+    "怜": 3899,
+    "思": 535,
+    "怡": 2740,
+    "急": 985,
+    "性": 175,
+    "怨": 4202,
+    "怪": 1587,
+    "怵": 4458,
+    "总": 409,
+    "恋": 2444,
+    "恍": 4323,
+    "恐": 828,
+    "恒": 1220,
+    "恕": 3803,
+    "恢": 1178,
+    "恨": 2812,
+    "恩": 995,
+    "恬": 3692,
+    "恭": 2621,
+    "息": 631,
+    "恰": 1359,
+    "恶": 868,
+    "恼": 3524,
+    "恽": 4349,
+    "悄": 3611,
+    "悉": 1156,
+    "悌": 4482,
+    "悔": 3649,
+    "悖": 3355,
+    "悚": 3579,
+    "悟": 1728,
+    "悠": 1714,
+    "患": 858,
+    "悦": 2777,
+    "您": 1223,
+    "悫": 4143,
+    "悬": 1603,
+    "悲": 2866,
+    "悼": 2805,
+    "情": 223,
+    "惊": 1440,
+    "惑": 2308,
+    "惕": 2304,
+    "惘": 3794,
+    "惜": 1657,
+    "惟": 2734,
+    "惠": 1330,
+    "惧": 1218,
+    "惨": 2307,
+    "惩": 3165,
+    "惪": 4203,
+    "惬": 2306,
+    "惯": 1231,
+    "惰": 3604,
+    "想": 394,
+    "惶": 3841,
+    "惹": 2999,
+    "惺": 4180,
+    "愁": 3822,
+    "愈": 2004,
+    "愉": 3092,
+    "意": 138,
+    "愔": 4376,
+    "愕": 3757,
+    "愚": 1975,
+    "感": 359,
+    "愤": 1816,
+    "愧": 4494,
+    "愿": 640,
+    "慈": 2066,
+    "慌": 3300,
+    "慎": 1755,
+    "慑": 3872,
+    "慕": 2522,
+    "慢": 1319,
+    "慧": 2607,
+    "慨": 1917,
+    "慰": 2746,
+    "慷": 3363,
+    "憍": 3737,
+    "憧": 3361,
+    "憨": 4093,
+    "憩": 3277,
+    "憬": 3362,
+    "憾": 1630,
+    "懂": 1446,
+    "懈": 3625,
+    "懋": 3006,
+    "懒": 2115,
+    "懔": 4168,
+    "懦": 4426,
+    "懿": 3032,
+    "戈": 749,
+    "戊": 2600,
+    "戌": 2573,
+    "戍": 3225,
+    "戎": 2815,
+    "戏": 820,
+    "成": 51,
+    "我": 166,
+    "戒": 1758,
+    "或": 68,
+    "战": 213,
+    "戚": 3132,
+    "戛": 4486,
+    "戟": 2546,
+    "截": 1033,
+    "戮": 3262,
+    "戴": 1513,
+    "户": 515,
+    "房": 633,
+    "所": 99,
+    "扁": 2498,
+    "扇": 2732,
+    "扈": 3128,
+    "扉": 3945,
+    "手": 173,
+    "才": 454,
+    "扎": 895,
+    "扑": 2551,
+    "打": 392,
+    "扔": 2068,
+    "托": 665,
+    "扣": 3078,
+    "执": 1058,
+    "扩": 1425,
+    "扫": 1336,
+    "扬": 1834,
+    "扭": 2159,
+    "扮": 2180,
+    "扯": 3252,
+    "扰": 1828,
+    "扳": 4162,
+    "扶": 2330,
+    "批": 846,
+    "扼": 4053,
+    "找": 781,
+    "承": 813,
+    "技": 719,
+    "抄": 1988,
+    "把": 463,
+    "抑": 1753,
+    "抒": 3467,
+    "抓": 1907,
+    "投": 568,
+    "抖": 3885,
+    "抗": 470,
+    "折": 1066,
+    "抚": 1764,
+    "抛": 1786,
+    "抢": 1241,
+    "护": 601,
+    "报": 237,
+    "披": 2679,
+    "抬": 2974,
+    "抱": 2205,
+    "抵": 1211,
+    "抹": 3617,
+    "押": 1473,
+    "抽": 1751,
+    "抿": 2208,
+    "拂": 2857,
+    "担": 688,
+    "拆": 2559,
+    "拇": 4106,
+    "拉": 157,
+    "拌": 4303,
+    "拍": 1006,
+    "拐": 2058,
+    "拒": 1041,
+    "拓": 2015,
+    "拔": 1812,
+    "拖": 1594,
+    "拘": 1460,
+    "招": 2030,
+    "拜": 1357,
+    "拟": 1386,
+    "拢": 4396,
+    "拥": 854,
+    "拦": 2137,
+    "拧": 4011,
+    "拨": 1469,
+    "择": 888,
+    "括": 401,
+    "拮": 3606,
+    "拯": 3257,
+    "拱": 2537,
+    "拳": 2547,
+    "拴": 1984,
+    "拷": 2199,
+    "拼": 1570,
+    "拾": 3356,
+    "拿": 775,
+    "持": 278,
+    "挂": 2257,
+    "指": 299,
+    "按": 1049,
+    "挑": 1548,
+    "挖": 2192,
+    "挚": 4198,
+    "挛": 3298,
+    "挝": 2877,
+    "挞": 4399,
+    "挟": 2050,
+    "挠": 4006,
+    "挡": 1665,
+    "挤": 1574,
+    "挥": 869,
+    "挪": 1048,
+    "挫": 4164,
+    "振": 1278,
+    "挽": 2161,
+    "捆": 3480,
+    "捉": 1856,
+    "捍": 4341,
+    "捏": 3906,
+    "捐": 1866,
+    "捕": 838,
+    "捞": 3279,
+    "损": 1005,
+    "捡": 4412,
+    "换": 742,
+    "捣": 3904,
+    "捧": 3578,
+    "捩": 3871,
+    "据": 267,
+    "捶": 1991,
+    "捷": 1624,
+    "授": 892,
+    "掉": 1428,
+    "掌": 1608,
+    "掏": 2060,
+    "排": 797,
+    "掖": 3274,
+    "掘": 2026,
+    "掛": 3924,
+    "掞": 3855,
+    "掠": 1656,
+    "探": 951,
+    "接": 247,
+    "控": 540,
+    "推": 424,
+    "掩": 1768,
+    "措": 1210,
+    "掰": 4324,
+    "掳": 3717,
+    "掷": 1464,
+    "掸": 3666,
+    "揆": 4478,
+    "揉": 4483,
+    "描": 1158,
+    "提": 130,
+    "插": 1966,
+    "揖": 3371,
+    "握": 1538,
+    "揩": 3988,
+    "揭": 2061,
+    "援": 876,
+    "搁": 1947,
+    "搅": 4302,
+    "搏": 3977,
+    "搔": 3998,
+    "搜": 1063,
+    "搞": 1903,
+    "搬": 1922,
+    "搭": 1520,
+    "携": 1093,
+    "摄": 809,
+    "摆": 1430,
+    "摇": 2133,
+    "摊": 3055,
+    "摔": 1846,
+    "摘": 2196,
+    "摧": 1267,
+    "摩": 851,
+    "摸": 1989,
+    "摹": 3722,
+    "撇": 4285,
+    "撑": 1963,
+    "撒": 1134,
+    "撕": 2113,
+    "撞": 994,
+    "撤": 1304,
+    "播": 622,
+    "撮": 4485,
+    "撰": 1954,
+    "撷": 4287,
+    "撼": 2102,
+    "擂": 3948,
+    "擅": 1640,
+    "操": 1228,
+    "擎": 2176,
+    "擒": 4261,
+    "擢": 2854,
+    "擦": 1813,
+    "攀": 1736,
+    "攒": 4121,
+    "攥": 4236,
+    "支": 450,
+    "收": 403,
+    "攸": 4082,
+    "改": 564,
+    "攻": 867,
+    "放": 361,
+    "政": 287,
+    "故": 523,
+    "效": 704,
+    "敌": 1486,
+    "敏": 1427,
+    "救": 763,
+    "敕": 3381,
+    "敖": 3240,
+    "教": 217,
+    "敛": 2139,
+    "敞": 2203,
+    "敢": 2143,
+    "散": 1045,
+    "敦": 1077,
+    "敬": 2435,
+    "数": 137,
+    "敲": 1388,
+    "整": 679,
+    "敷": 3153,
+    "文": 148,
+    "斋": 3072,
+    "斌": 3148,
+    "斐": 1654,
+    "斑": 1824,
+    "斗": 1411,
+    "料": 842,
+    "斛": 2976,
+    "斜": 1268,
+    "斝": 4293,
+    "斟": 4292,
+    "斡": 4367,
+    "斤": 1635,
+    "斥": 2966,
+    "斧": 3103,
+    "斩": 2842,
+    "断": 647,
+    "斯": 46,
+    "新": 145,
+    "方": 58,
+    "於": 3071,
+    "施": 678,
+    "旁": 1503,
+    "旅": 164,
+    "旋": 943,
+    "旌": 4214,
+    "族": 783,
+    "旗": 1321,
+    "无": 318,
+    "既": 1117,
+    "日": 146,
+    "旦": 1522,
+    "旧": 1266,
+    "旨": 1539,
+    "早": 544,
+    "旬": 2168,
+    "旭": 3190,
+    "旱": 1671,
+    "时": 24,
+    "旺": 1353,
+    "昀": 3402,
+    "昂": 1292,
+    "昆": 1575,
+    "昊": 3005,
+    "昌": 2360,
+    "明": 215,
+    "昏": 3120,
+    "易": 513,
+    "昔": 2240,
+    "昕": 3789,
+    "昙": 3577,
+    "星": 304,
+    "映": 2105,
+    "春": 1825,
+    "昨": 1443,
+    "昭": 2447,
+    "是": 7,
+    "昱": 3870,
+    "昴": 4347,
+    "昵": 4336,
+    "昼": 1689,
+    "显": 430,
+    "晁": 3374,
+    "晃": 2764,
+    "晋": 2372,
+    "晏": 3082,
+    "晒": 3399,
+    "晓": 1459,
+    "晔": 3878,
+    "晕": 2874,
+    "晖": 2741,
+    "晚": 658,
+    "晟": 3294,
+    "晤": 3030,
+    "晦": 4451,
+    "晨": 1713,
+    "普": 398,
+    "景": 581,
+    "晰": 1797,
+    "晴": 2111,
+    "晶": 2449,
+    "智": 909,
+    "暂": 1112,
+    "暄": 4176,
+    "暎": 4167,
+    "暑": 1763,
+    "暖": 1389,
+    "暗": 1194,
+    "暝": 4003,
+    "暨": 3404,
+    "暮": 4207,
+    "暴": 664,
+    "暹": 2301,
+    "曙": 3752,
+    "曜": 3965,
+    "曝": 1537,
+    "曰": 3448,
+    "曲": 1264,
+    "更": 111,
+    "曷": 3245,
+    "曹": 2428,
+    "曼": 853,
+    "曾": 452,
+    "替": 1474,
+    "最": 81,
+    "月": 139,
+    "有": 8,
+    "朋": 1511,
+    "服": 412,
+    "朔": 2670,
+    "朕": 3938,
+    "朗": 978,
+    "望": 577,
+    "朝": 1090,
+    "期": 206,
+    "木": 758,
+    "未": 365,
+    "末": 1317,
+    "本": 153,
+    "札": 2666,
+    "术": 511,
+    "朱": 1580,
+    "朴": 2695,
+    "朵": 3117,
+    "机": 142,
+    "朽": 3999,
+    "杀": 1108,
+    "杂": 948,
+    "权": 500,
+    "杆": 2232,
+    "杉": 1421,
+    "李": 1182,
+    "杏": 3756,
+    "材": 1517,
+    "村": 1204,
+    "杓": 3986,
+    "杖": 2743,
+    "杙": 4316,
+    "杜": 1904,
+    "杞": 2086,
+    "束": 746,
+    "杠": 3619,
+    "条": 571,
+    "来": 43,
+    "杨": 2357,
+    "杭": 2468,
+    "杯": 1596,
+    "杰": 731,
+    "杵": 4159,
+    "杻": 4449,
+    "松": 754,
+    "板": 923,
+    "极": 484,
+    "构": 590,
+    "枋": 3633,
+    "析": 2403,
+    "枕": 3271,
+    "林": 486,
+    "枚": 1643,
+    "果": 161,
+    "枝": 1898,
+    "枞": 4100,
+    "枢": 2163,
+    "枣": 3034,
+    "枥": 4163,
+    "枪": 1127,
+    "枫": 3382,
+    "枭": 4335,
+    "枯": 2832,
+    "架": 792,
+    "柃": 3177,
+    "柄": 2470,
+    "柏": 1122,
+    "某": 596,
+    "柑": 3035,
+    "染": 561,
+    "柔": 1320,
+    "柚": 3410,
+    "柜": 1910,
+    "柝": 3571,
+    "柞": 4257,
+    "柠": 3110,
+    "查": 324,
+    "柩": 3867,
+    "柬": 1370,
+    "柯": 2122,
+    "柰": 3766,
+    "柱": 1973,
+    "柳": 2377,
+    "柴": 2523,
+    "査": 3504,
+    "柽": 2896,
+    "柿": 2170,
+    "栀": 4086,
+    "栃": 3913,
+    "栅": 2227,
+    "标": 339,
+    "栈": 3432,
+    "栉": 3466,
+    "栋": 1667,
+    "栎": 4317,
+    "栏": 1835,
+    "树": 761,
+    "栓": 3585,
+    "栖": 2412,
+    "栗": 2727,
+    "校": 748,
+    "栢": 4220,
+    "栩": 3142,
+    "株": 2273,
+    "样": 226,
+    "核": 617,
+    "根": 481,
+    "栻": 3767,
+    "格": 250,
+    "栽": 2961,
+    "栾": 3538,
+    "桂": 2437,
+    "桃": 2189,
+    "桄": 4318,
+    "框": 1447,
+    "案": 499,
+    "桉": 3046,
+    "桌": 2934,
+    "桐": 2452,
+    "桑": 1396,
+    "桓": 3044,
+    "桔": 2545,
+    "桝": 4020,
+    "桡": 3946,
+    "桢": 3774,
+    "档": 1265,
+    "桤": 2987,
+    "桥": 871,
+    "桦": 3056,
+    "桧": 3613,
+    "桨": 2964,
+    "桩": 2228,
+    "桫": 3433,
+    "桶": 1171,
+    "梁": 1552,
+    "梅": 1527,
+    "梓": 2975,
+    "梗": 2802,
+    "梢": 2995,
+    "梦": 1250,
+    "梧": 3211,
+    "梨": 2567,
+    "梭": 2623,
+    "梯": 2041,
+    "械": 2465,
+    "梳": 3820,
+    "梵": 1039,
+    "梾": 3162,
+    "检": 702,
+    "棁": 4427,
+    "棉": 3070,
+    "棋": 2383,
+    "棍": 2190,
+    "棒": 1510,
+    "棕": 1531,
+    "棘": 2263,
+    "棚": 3329,
+    "棠": 2780,
+    "棣": 3342,
+    "棨": 4475,
+    "森": 852,
+    "棱": 2221,
+    "棵": 1670,
+    "棺": 2347,
+    "椅": 1506,
+    "椋": 3875,
+    "植": 824,
+    "椎": 2810,
+    "椒": 2524,
+    "検": 3970,
+    "椤": 3223,
+    "椭": 2804,
+    "椰": 1981,
+    "椹": 3738,
+    "椿": 3125,
+    "楚": 1392,
+    "楝": 3018,
+    "楞": 3556,
+    "楠": 2758,
+    "楮": 4275,
+    "楷": 3765,
+    "楸": 3763,
+    "楼": 1318,
+    "概": 1444,
+    "榄": 1184,
+    "榆": 2633,
+    "榈": 1716,
+    "榉": 3864,
+    "榔": 3659,
+    "榕": 2876,
+    "榖": 3954,
+    "榙": 4245,
+    "榛": 2151,
+    "榜": 1283,
+    "榧": 3060,
+    "榨": 2238,
+    "榭": 4457,
+    "榴": 2290,
+    "榻": 1512,
+    "槐": 2700,
+    "槚": 4354,
+    "槛": 4182,
+    "槟": 2723,
+    "槭": 2509,
+    "槱": 4296,
+    "槲": 3740,
+    "槽": 3135,
+    "樊": 3025,
+    "樟": 2531,
+    "模": 605,
+    "樨": 4415,
+    "横": 1225,
+    "樱": 1833,
+    "樵": 4387,
+    "樽": 2889,
+    "樾": 4338,
+    "橄": 1183,
+    "橇": 2024,
+    "橐": 3420,
+    "橘": 2730,
+    "橙": 2611,
+    "橡": 2658,
+    "橱": 2319,
+    "檀": 3413,
+    "檐": 3158,
+    "檗": 2539,
+    "檬": 2916,
+    "欠": 2888,
+    "次": 168,
+    "欢": 796,
+    "欣": 1602,
+    "欧": 396,
+    "欲": 2217,
+    "欺": 2075,
+    "款": 1245,
+    "歆": 3848,
+    "歇": 1855,
+    "歉": 1582,
+    "歌": 949,
+    "歙": 3183,
+    "止": 681,
+    "正": 193,
+    "此": 191,
+    "步": 485,
+    "武": 992,
+    "歧": 1718,
+    "歪": 1902,
+    "歹": 3900,
+    "死": 649,
+    "歼": 3343,
+    "殃": 3630,
+    "殆": 3791,
+    "殉": 2779,
+    "殊": 1309,
+    "残": 1700,
+    "殒": 4067,
+    "殖": 1302,
+    "殴": 3422,
+    "段": 772,
+    "殷": 3312,
+    "殿": 1079,
+    "毁": 918,
+    "毅": 2630,
+    "毋": 3412,
+    "母": 637,
+    "每": 255,
+    "毒": 582,
+    "毓": 3634,
+    "比": 102,
+    "毕": 1065,
+    "毗": 2816,
+    "毙": 3033,
+    "毛": 801,
+    "毫": 1307,
+    "毯": 2152,
+    "氏": 1435,
+    "民": 253,
+    "氓": 3731,
+    "气": 300,
+    "氖": 4371,
+    "氙": 4456,
+    "氛": 1776,
+    "氟": 2648,
+    "氢": 1696,
+    "氦": 4397,
+    "氧": 1790,
+    "氨": 1940,
+    "氮": 1576,
+    "氯": 2276,
+    "氰": 1840,
+    "水": 194,
+    "永": 1167,
+    "汀": 2284,
+    "汁": 2239,
+    "求": 444,
+    "汇": 1208,
+    "汉": 1193,
+    "汐": 2292,
+    "汕": 3728,
+    "汗": 1020,
+    "汛": 3208,
+    "汜": 4036,
+    "汝": 2485,
+    "汞": 3650,
+    "江": 2348,
+    "池": 1600,
+    "污": 1031,
+    "汤": 1139,
+    "汪": 2560,
+    "汰": 2010,
+    "汲": 3694,
+    "汴": 3517,
+    "汶": 3213,
+    "汹": 4071,
+    "汽": 672,
+    "汾": 3249,
+    "沁": 2898,
+    "沂": 2721,
+    "沃": 1529,
+    "沅": 2948,
+    "沆": 3557,
+    "沈": 2471,
+    "沉": 1395,
+    "沌": 3836,
+    "沐": 4075,
+    "沔": 3886,
+    "沙": 685,
+    "沛": 2919,
+    "沟": 2406,
+    "没": 152,
+    "沤": 4210,
+    "沥": 4297,
+    "沦": 3434,
+    "沧": 2592,
+    "沪": 2711,
+    "沫": 1731,
+    "沭": 4391,
+    "沮": 2087,
+    "河": 433,
+    "沸": 3241,
+    "油": 550,
+    "治": 451,
+    "沼": 1897,
+    "沽": 3003,
+    "沾": 3632,
+    "沿": 791,
+    "泄": 1133,
+    "泉": 2297,
+    "泊": 1468,
+    "泌": 2849,
+    "泓": 4113,
+    "法": 55,
+    "泗": 3246,
+    "泛": 1284,
+    "泠": 3844,
+    "泡": 980,
+    "波": 475,
+    "泣": 3285,
+    "泥": 1868,
+    "注": 687,
+    "泪": 2266,
+    "泮": 3918,
+    "泯": 3972,
+    "泰": 1019,
+    "泱": 3729,
+    "泳": 1436,
+    "泵": 1892,
+    "泷": 4237,
+    "泸": 3805,
+    "泻": 2083,
+    "泼": 3917,
+    "泽": 1227,
+    "泾": 3023,
+    "洁": 1821,
+    "洋": 594,
+    "洐": 4444,
+    "洒": 2036,
+    "洗": 2140,
+    "洙": 3943,
+    "洛": 419,
+    "洞": 721,
+    "洣": 3976,
+    "津": 2135,
+    "洪": 1368,
+    "洮": 3706,
+    "洱": 3449,
+    "洲": 296,
+    "洵": 3824,
+    "活": 216,
+    "洽": 2121,
+    "派": 1100,
+    "流": 303,
+    "浅": 1948,
+    "浆": 2776,
+    "浇": 3500,
+    "浉": 3950,
+    "浊": 3292,
+    "测": 497,
+    "济": 661,
+    "浏": 2578,
+    "浑": 2935,
+    "浓": 1191,
+    "浔": 3847,
+    "浙": 2364,
+    "浚": 4119,
+    "浞": 4417,
+    "浦": 2415,
+    "浩": 2093,
+    "浪": 1255,
+    "浮": 1467,
+    "浴": 2282,
+    "海": 200,
+    "浸": 4360,
+    "涂": 1261,
+    "涅": 2230,
+    "消": 416,
+    "涉": 1064,
+    "涌": 2262,
+    "涓": 4358,
+    "涛": 1791,
+    "涝": 3982,
+    "涞": 3753,
+    "涟": 3584,
+    "涡": 2872,
+    "润": 1811,
+    "涧": 3133,
+    "涨": 3106,
+    "涩": 2930,
+    "涪": 3553,
+    "涯": 1832,
+    "液": 946,
+    "涵": 1704,
+    "涸": 1772,
+    "涿": 4439,
+    "淀": 2698,
+    "淄": 2817,
+    "淅": 3680,
+    "淆": 3050,
+    "淇": 3642,
+    "淋": 3042,
+    "淌": 2027,
+    "淑": 3206,
+    "淖": 3742,
+    "淘": 2009,
+    "淞": 4466,
+    "淡": 2346,
+    "淤": 4377,
+    "淫": 2787,
+    "淮": 2581,
+    "淯": 3651,
+    "深": 1021,
+    "淳": 2762,
+    "混": 1647,
+    "淹": 2076,
+    "添": 1780,
+    "淼": 4127,
+    "清": 805,
+    "渊": 1986,
+    "渌": 3843,
+    "渎": 3554,
+    "渐": 1955,
+    "渔": 1887,
+    "渗": 3416,
+    "渚": 3798,
+    "渝": 2925,
+    "渠": 2675,
+    "渡": 1551,
+    "渣": 3503,
+    "渤": 3490,
+    "渥": 2677,
+    "温": 831,
+    "渭": 3218,
+    "港": 826,
+    "渲": 3779,
+    "渴": 3695,
+    "游": 169,
+    "渺": 4332,
+    "湄": 2946,
+    "湍": 2174,
+    "湖": 883,
+    "湘": 2606,
+    "湛": 3450,
+    "湜": 4129,
+    "湟": 4433,
+    "湫": 4330,
+    "湳": 3724,
+    "湾": 744,
+    "湿": 1483,
+    "溃": 3116,
+    "溅": 4156,
+    "溉": 3424,
+    "源": 572,
+    "準": 2870,
+    "溜": 1142,
+    "溞": 3330,
+    "溢": 1893,
+    "溥": 3220,
+    "溧": 2871,
+    "溪": 1733,
+    "溯": 1238,
+    "溲": 3238,
+    "溴": 2902,
+    "溶": 2443,
+    "溺": 3372,
+    "滁": 3544,
+    "滇": 2594,
+    "滋": 1653,
+    "滑": 559,
+    "滕": 1842,
+    "滚": 1819,
+    "滞": 3318,
+    "满": 692,
+    "滤": 1738,
+    "滥": 1993,
+    "滨": 2365,
+    "滩": 1360,
+    "滴": 2884,
+    "漂": 1952,
+    "漆": 2028,
+    "漏": 1013,
+    "演": 438,
+    "漕": 3440,
+    "漠": 1765,
+    "漩": 3549,
+    "漪": 3447,
+    "漫": 1150,
+    "漳": 2785,
+    "漾": 3832,
+    "潇": 4208,
+    "潘": 2417,
+    "潜": 527,
+    "潞": 3541,
+    "潢": 3539,
+    "潭": 2556,
+    "潮": 1462,
+    "澄": 3062,
+    "澎": 2859,
+    "澜": 2793,
+    "澡": 3689,
+    "澥": 4273,
+    "澧": 3405,
+    "澳": 662,
+    "激": 790,
+    "濂": 3305,
+    "濉": 4491,
+    "濑": 2903,
+    "濒": 3244,
+    "濠": 3980,
+    "濡": 3927,
+    "濮": 4019,
+    "濯": 4382,
+    "瀑": 1145,
+    "瀚": 2094,
+    "灌": 2705,
+    "灏": 3647,
+    "灞": 4158,
+    "火": 464,
+    "灭": 2016,
+    "灯": 881,
+    "灰": 2390,
+    "灵": 1125,
+    "灶": 3263,
+    "灼": 2955,
+    "灾": 1472,
+    "灿": 3387,
+    "炀": 3136,
+    "炉": 2019,
+    "炊": 4001,
+    "炎": 1296,
+    "炒": 3396,
+    "炔": 3712,
+    "炕": 3817,
+    "炜": 4455,
+    "炫": 4383,
+    "炬": 3309,
+    "炭": 2021,
+    "炮": 1965,
+    "炸": 767,
+    "点": 225,
+    "炼": 1619,
+    "炽": 3173,
+    "烁": 3648,
+    "烂": 3242,
+    "烃": 2774,
+    "烈": 1022,
+    "烘": 3400,
+    "烙": 4409,
+    "烛": 2724,
+    "烟": 2057,
+    "烤": 1568,
+    "烦": 1327,
+    "烧": 1310,
+    "烨": 3368,
+    "烫": 2333,
+    "热": 434,
+    "烯": 2773,
+    "烷": 2665,
+    "烹": 1633,
+    "烺": 4362,
+    "烽": 3102,
+    "焉": 3592,
+    "焊": 3094,
+    "焕": 3484,
+    "焚": 3920,
+    "焦": 747,
+    "焮": 3915,
+    "焰": 1946,
+    "焱": 4097,
+    "然": 114,
+    "煌": 3016,
+    "煎": 3837,
+    "煜": 3624,
+    "煤": 1303,
+    "照": 563,
+    "煮": 2904,
+    "煲": 4024,
+    "熄": 3478,
+    "熈": 4117,
+    "熊": 1470,
+    "熏": 4331,
+    "熔": 2892,
+    "熙": 2402,
+    "熟": 953,
+    "熠": 4008,
+    "熨": 2269,
+    "熬": 3439,
+    "燃": 1407,
+    "燕": 2501,
+    "燥": 1925,
+    "燮": 3310,
+    "爆": 938,
+    "爪": 1350,
+    "爬": 1734,
+    "爱": 506,
+    "爵": 1383,
+    "父": 1189,
+    "爷": 3038,
+    "爸": 2577,
+    "爹": 4379,
+    "爽": 3147,
+    "牁": 4430,
+    "牂": 4429,
+    "片": 469,
+    "版": 834,
+    "牌": 1272,
+    "牕": 4190,
+    "牙": 620,
+    "牛": 1172,
+    "牟": 3166,
+    "牡": 2755,
+    "牢": 2328,
+    "牦": 3445,
+    "牧": 981,
+    "物": 64,
+    "牲": 2299,
+    "牵": 1961,
+    "特": 95,
+    "牺": 2298,
+    "牻": 3333,
+    "犀": 2840,
+    "犁": 2735,
+    "犍": 3386,
+    "犒": 1774,
+    "犬": 1789,
+    "犯": 962,
+    "状": 437,
+    "犷": 4157,
+    "犹": 1164,
+    "狂": 2018,
+    "狄": 2051,
+    "狐": 2729,
+    "狗": 1964,
+    "狙": 3693,
+    "狠": 3776,
+    "狡": 3340,
+    "狩": 1895,
+    "独": 529,
+    "狭": 2212,
+    "狮": 1200,
+    "狯": 4228,
+    "狱": 1173,
+    "狸": 2549,
+    "狼": 1406,
+    "猁": 2064,
+    "猄": 4394,
+    "猎": 1061,
+    "猕": 3635,
+    "猗": 3678,
+    "猛": 1213,
+    "猜": 2138,
+    "猝": 4459,
+    "猞": 2063,
+    "猩": 1452,
+    "猪": 1777,
+    "猫": 990,
+    "猬": 3207,
+    "献": 1577,
+    "猴": 2219,
+    "猿": 1960,
+    "獐": 3087,
+    "獭": 1909,
+    "獴": 4146,
+    "玄": 2469,
+    "率": 788,
+    "玉": 2169,
+    "王": 532,
+    "玎": 4390,
+    "玖": 3261,
+    "玛": 1471,
+    "玠": 4045,
+    "玩": 988,
+    "玫": 2736,
+    "玭": 4350,
+    "玮": 3200,
+    "环": 705,
+    "现": 74,
+    "玲": 3272,
+    "玶": 4250,
+    "玹": 4154,
+    "玻": 1378,
+    "珀": 1419,
+    "珂": 3317,
+    "珈": 4465,
+    "珊": 2420,
+    "珍": 1480,
+    "珐": 3940,
+    "珑": 3816,
+    "珙": 3345,
+    "珞": 3733,
+    "珠": 2054,
+    "珧": 3461,
+    "珩": 3542,
+    "班": 489,
+    "珰": 4364,
+    "球": 118,
+    "琅": 3113,
+    "理": 156,
+    "琉": 2542,
+    "琊": 3838,
+    "琐": 4064,
+    "琚": 3718,
+    "琛": 3764,
+    "琢": 3502,
+    "琥": 3566,
+    "琦": 2918,
+    "琨": 3631,
+    "琪": 2921,
+    "琬": 4392,
+    "琮": 3197,
+    "琰": 3388,
+    "琳": 1921,
+    "琴": 1243,
+    "琵": 3675,
+    "琶": 4077,
+    "琼": 2691,
+    "瑀": 3641,
+    "瑙": 2184,
+    "瑚": 2441,
+    "瑛": 3187,
+    "瑜": 2897,
+    "瑞": 716,
+    "瑟": 1639,
+    "瑭": 4493,
+    "瑮": 3973,
+    "瑰": 2977,
+    "瑶": 3408,
+    "瑾": 3616,
+    "璀": 2073,
+    "璃": 1379,
+    "璆": 4189,
+    "璇": 3746,
+    "璋": 3302,
+    "璎": 4205,
+    "璜": 3479,
+    "璧": 3773,
+    "璨": 2074,
+    "璹": 4171,
+    "瓘": 4357,
+    "瓛": 4258,
+    "瓜": 1154,
+    "瓢": 3339,
+    "瓣": 1915,
+    "瓦": 634,
+    "瓮": 4269,
+    "瓯": 4012,
+    "瓶": 1565,
+    "瓷": 2516,
+    "甄": 2173,
+    "甘": 2177,
+    "甚": 597,
+    "甜": 2243,
+    "生": 40,
+    "甥": 3109,
+    "用": 53,
+    "甩": 2181,
+    "甫": 2553,
+    "甬": 3077,
+    "田": 1783,
+    "由": 176,
+    "甲": 1644,
+    "申": 1285,
+    "电": 197,
+    "男": 890,
+    "甸": 2421,
+    "町": 2359,
+    "画": 920,
+    "甾": 3984,
+    "畅": 1554,
+    "畋": 4088,
+    "界": 262,
+    "畏": 3124,
+    "畔": 2370,
+    "留": 630,
+    "畜": 2322,
+    "略": 830,
+    "番": 2110,
+    "畲": 4356,
+    "畴": 2931,
+    "畸": 2873,
+    "畿": 3089,
+    "疃": 3682,
+    "疆": 1625,
+    "疍": 3812,
+    "疏": 1677,
+    "疑": 914,
+    "疗": 886,
+    "疟": 1859,
+    "疡": 3842,
+    "疣": 3319,
+    "疤": 4402,
+    "疫": 874,
+    "疮": 3471,
+    "疯": 2332,
+    "疲": 1849,
+    "疹": 4288,
+    "疼": 1418,
+    "疽": 2162,
+    "疾": 893,
+    "病": 367,
+    "症": 859,
+    "痉": 3685,
+    "痒": 3622,
+    "痕": 2819,
+    "痘": 4469,
+    "痛": 1175,
+    "痢": 3891,
+    "痪": 4240,
+    "痴": 2120,
+    "痼": 4425,
+    "瘠": 4435,
+    "瘤": 2128,
+    "瘦": 3306,
+    "瘫": 4239,
+    "瘰": 4225,
+    "瘾": 3435,
+    "瘿": 3431,
+    "癌": 1433,
+    "癸": 2631,
+    "登": 739,
+    "白": 641,
+    "百": 1095,
+    "皂": 3011,
+    "的": 4,
+    "皆": 2395,
+    "皇": 806,
+    "皋": 3534,
+    "皎": 3669,
+    "皓": 4098,
+    "皕": 4300,
+    "皖": 3301,
+    "皮": 759,
+    "皱": 2808,
+    "皿": 2676,
+    "盂": 3953,
+    "盆": 2118,
+    "盈": 2267,
+    "益": 937,
+    "盎": 1760,
+    "盏": 4407,
+    "盐": 2275,
+    "监": 724,
+    "盒": 2782,
+    "盔": 2172,
+    "盖": 752,
+    "盗": 1488,
+    "盘": 1195,
+    "盛": 1046,
+    "盟": 632,
+    "目": 155,
+    "盱": 3825,
+    "盲": 1886,
+    "直": 326,
+    "相": 189,
+    "盼": 3928,
+    "盾": 1885,
+    "省": 862,
+    "眉": 2644,
+    "看": 254,
+    "県": 4031,
+    "眙": 3826,
+    "真": 707,
+    "眠": 1932,
+    "眨": 1852,
+    "眩": 4378,
+    "眶": 2268,
+    "眷": 3687,
+    "眼": 845,
+    "着": 174,
+    "睁": 4081,
+    "睛": 1636,
+    "睡": 1128,
+    "睢": 3933,
+    "督": 972,
+    "睦": 3079,
+    "睫": 3813,
+    "睹": 3802,
+    "睾": 3086,
+    "睿": 3637,
+    "瞄": 3823,
+    "瞎": 3861,
+    "瞧": 3790,
+    "瞩": 1958,
+    "瞬": 3546,
+    "瞭": 3468,
+    "瞰": 2043,
+    "瞳": 3874,
+    "瞻": 3620,
+    "矗": 3969,
+    "矛": 1544,
+    "矢": 3066,
+    "矣": 3175,
+    "知": 279,
+    "矩": 1959,
+    "矫": 3359,
+    "短": 784,
+    "矮": 2738,
+    "石": 520,
+    "矶": 1422,
+    "矾": 2858,
+    "矿": 2410,
+    "砀": 4140,
+    "码": 768,
+    "砂": 2274,
+    "砌": 3338,
+    "砍": 3437,
+    "研": 614,
+    "砖": 2713,
+    "砗": 3561,
+    "砝": 4320,
+    "砥": 4386,
+    "破": 723,
+    "砵": 4144,
+    "砷": 3588,
+    "砸": 3614,
+    "砻": 3495,
+    "砾": 3956,
+    "础": 1380,
+    "硅": 2856,
+    "硎": 4281,
+    "硒": 3735,
+    "硕": 2034,
+    "硖": 4280,
+    "硝": 2638,
+    "硫": 2211,
+    "硬": 1391,
+    "确": 369,
+    "硼": 3601,
+    "碍": 1237,
+    "碎": 1152,
+    "碑": 1945,
+    "碗": 3228,
+    "碘": 2467,
+    "碟": 3178,
+    "碧": 2783,
+    "碰": 1854,
+    "碱": 2316,
+    "碲": 3129,
+    "碳": 1779,
+    "碾": 1992,
+    "磁": 1123,
+    "磅": 1333,
+    "磐": 2991,
+    "磡": 3587,
+    "磨": 2293,
+    "磲": 3562,
+    "磷": 2210,
+    "礁": 2258,
+    "示": 183,
+    "礼": 1206,
+    "社": 683,
+    "祀": 2643,
+    "祁": 3258,
+    "祇": 2283,
+    "祈": 2336,
+    "祉": 3793,
+    "祋": 3961,
+    "祕": 3391,
+    "祖": 1364,
+    "祗": 3569,
+    "祚": 3323,
+    "祜": 4137,
+    "祝": 1305,
+    "神": 411,
+    "祠": 2503,
+    "祥": 2451,
+    "票": 422,
+    "祭": 1747,
+    "祯": 2637,
+    "祷": 2337,
+    "祸": 1316,
+    "祺": 3141,
+    "禀": 4327,
+    "禁": 1017,
+    "禄": 2077,
+    "禅": 2657,
+    "福": 922,
+    "禑": 4126,
+    "禛": 4161,
+    "禧": 2936,
+    "禹": 2867,
+    "禺": 3537,
+    "离": 439,
+    "禽": 1086,
+    "禾": 2373,
+    "秀": 1363,
+    "私": 866,
+    "秃": 1631,
+    "秆": 3145,
+    "秉": 3014,
+    "秋": 2396,
+    "种": 69,
+    "科": 258,
+    "秒": 1367,
+    "秘": 1103,
+    "租": 1135,
+    "秤": 3661,
+    "秦": 2391,
+    "秧": 4411,
+    "秩": 2246,
+    "积": 823,
+    "称": 135,
+    "移": 606,
+    "秽": 4487,
+    "稀": 1144,
+    "程": 293,
+    "稍": 1290,
+    "税": 816,
+    "稔": 3901,
+    "稗": 3545,
+    "稚": 2656,
+    "稠": 3150,
+    "稣": 2654,
+    "稳": 1114,
+    "稷": 3284,
+    "稻": 2090,
+    "稼": 4309,
+    "稽": 3054,
+    "稿": 2693,
+    "穆": 1322,
+    "穗": 2454,
+    "穴": 1219,
+    "究": 615,
+    "穷": 1827,
+    "穹": 3442,
+    "空": 208,
+    "穿": 696,
+    "突": 1016,
+    "窃": 1882,
+    "窄": 2213,
+    "窑": 1567,
+    "窗": 872,
+    "窜": 3710,
+    "窝": 2855,
+    "窟": 2927,
+    "窥": 3531,
+    "窦": 3017,
+    "竈": 3860,
+    "立": 280,
+    "竖": 3009,
+    "站": 395,
+    "竞": 891,
+    "竟": 1807,
+    "章": 2130,
+    "竣": 3996,
+    "童": 970,
+    "竭": 1850,
+    "端": 960,
+    "竹": 1920,
+    "竺": 2791,
+    "竿": 3299,
+    "笃": 3121,
+    "笄": 4428,
+    "笆": 4056,
+    "笋": 2971,
+    "笏": 3540,
+    "笑": 1741,
+    "笔": 1001,
+    "笙": 2801,
+    "笛": 2938,
+    "笠": 3019,
+    "符": 1203,
+    "笨": 1982,
+    "第": 167,
+    "笮": 4242,
+    "笼": 2504,
+    "等": 252,
+    "筋": 2280,
+    "筐": 3828,
+    "筑": 926,
+    "筒": 2286,
+    "答": 1487,
+    "策": 1082,
+    "筛": 3255,
+    "筮": 4027,
+    "筱": 3409,
+    "筷": 3942,
+    "筹": 2818,
+    "签": 557,
+    "简": 802,
+    "箕": 3596,
+    "算": 626,
+    "管": 220,
+    "箨": 4472,
+    "箩": 3827,
+    "箬": 3239,
+    "箭": 1558,
+    "箱": 935,
+    "箴": 3580,
+    "篇": 1752,
+    "篡": 2191,
+    "篦": 4363,
+    "篮": 1545,
+    "篱": 3407,
+    "篷": 2182,
+    "簇": 3311,
+    "簧": 3320,
+    "簪": 3806,
+    "簸": 4432,
+    "簿": 4116,
+    "籁": 4108,
+    "籍": 1384,
+    "米": 413,
+    "籴": 4076,
+    "类": 182,
+    "籽": 3063,
+    "粉": 1698,
+    "粒": 1354,
+    "粗": 1479,
+    "粘": 1729,
+    "粟": 2315,
+    "粤": 2790,
+    "粥": 3716,
+    "粪": 3446,
+    "粮": 2088,
+    "粹": 2256,
+    "精": 623,
+    "糊": 2126,
+    "糕": 1721,
+    "糖": 1402,
+    "糙": 2885,
+    "糟": 1720,
+    "糠": 4145,
+    "糯": 3668,
+    "系": 286,
+    "紊": 1967,
+    "紑": 4175,
+    "素": 689,
+    "索": 675,
+    "紧": 855,
+    "紫": 2362,
+    "累": 2459,
+    "綖": 4251,
+    "綦": 4193,
+    "緁": 4030,
+    "縻": 3000,
+    "繁": 1221,
+    "繄": 3935,
+    "纂": 3064,
+    "纠": 2869,
+    "红": 1126,
+    "纤": 2722,
+    "约": 233,
+    "级": 492,
+    "纪": 256,
+    "纬": 3288,
+    "纭": 3341,
+    "纮": 4022,
+    "纯": 2255,
+    "纱": 2062,
+    "纲": 2409,
+    "纳": 406,
+    "纵": 1927,
+    "纶": 3004,
+    "纷": 1595,
+    "纸": 1289,
+    "纹": 2079,
+    "纺": 2652,
+    "纻": 3876,
+    "纽": 1010,
+    "线": 381,
+    "绂": 3941,
+    "练": 860,
+    "组": 273,
+    "绅": 3227,
+    "细": 473,
+    "织": 460,
+    "终": 800,
+    "绍": 2386,
+    "绎": 3755,
+    "经": 89,
+    "绑": 2694,
+    "绒": 2585,
+    "结": 338,
+    "绕": 1009,
+    "绘": 1277,
+    "给": 526,
+    "绚": 4344,
+    "绛": 2981,
+    "络": 1188,
+    "绝": 674,
+    "绞": 4120,
+    "统": 274,
+    "绢": 3711,
+    "绣": 2571,
+    "绥": 2660,
+    "继": 887,
+    "绩": 1155,
+    "绪": 1652,
+    "绫": 4337,
+    "续": 504,
+    "绮": 3978,
+    "绯": 3555,
+    "绰": 3520,
+    "绳": 2554,
+    "维": 327,
+    "绵": 2014,
+    "绶": 3444,
+    "绸": 2096,
+    "综": 2022,
+    "绿": 1453,
+    "缀": 3575,
+    "缅": 2425,
+    "缆": 1553,
+    "缇": 3607,
+    "缉": 2188,
+    "缓": 1404,
+    "缔": 3334,
+    "缕": 2907,
+    "编": 1030,
+    "缘": 1339,
+    "缙": 4110,
+    "缚": 3406,
+    "缜": 3995,
+    "缝": 1591,
+    "缠": 1739,
+    "缨": 3423,
+    "缩": 1800,
+    "缪": 1944,
+    "缭": 1979,
+    "缮": 3393,
+    "缴": 2145,
+    "缵": 4128,
+    "缸": 3690,
+    "缺": 708,
+    "罂": 2314,
+    "罄": 1919,
+    "罐": 2928,
+    "网": 356,
+    "罔": 3992,
+    "罕": 1588,
+    "罗": 165,
+    "罘": 4254,
+    "罚": 1541,
+    "罢": 2728,
+    "罩": 3464,
+    "罪": 912,
+    "置": 709,
+    "署": 936,
+    "罹": 3688,
+    "羁": 2839,
+    "羊": 1242,
+    "羌": 2685,
+    "美": 144,
+    "羚": 2108,
+    "羞": 3599,
+    "羟": 3119,
+    "羡": 4474,
+    "群": 344,
+    "羧": 3214,
+    "羯": 4105,
+    "羰": 3797,
+    "羲": 3326,
+    "羽": 1666,
+    "翁": 2424,
+    "翃": 4461,
+    "翅": 1994,
+    "翊": 3013,
+    "翌": 3008,
+    "翎": 3684,
+    "翔": 2194,
+    "翘": 3392,
+    "翟": 3354,
+    "翠": 2668,
+    "翡": 3182,
+    "翥": 3829,
+    "翰": 720,
+    "翱": 2193,
+    "翻": 1076,
+    "翼": 2385,
+    "耀": 2530,
+    "老": 578,
+    "考": 611,
+    "者": 85,
+    "耆": 2706,
+    "而": 84,
+    "���": 4136,
+    "耐": 1180,
+    "耕": 2739,
+    "耗": 2001,
+    "耧": 3316,
+    "耳": 690,
+    "耶": 996,
+    "耸": 2082,
+    "耻": 4221,
+    "耽": 4130,
+    "耿": 3051,
+    "聂": 3139,
+    "聊": 1672,
+    "聋": 3164,
+    "职": 875,
+    "联": 234,
+    "聘": 2101,
+    "聚": 1092,
+    "聪": 1757,
+    "聿": 3937,
+    "肃": 2463,
+    "肄": 3289,
+    "肆": 2114,
+    "肇": 2970,
+    "肉": 2209,
+    "肋": 2861,
+    "肌": 2529,
+    "肖": 2565,
+    "肘": 3681,
+    "肚": 2218,
+    "肛": 2906,
+    "肝": 1746,
+    "肟": 3703,
+    "肠": 1775,
+    "股": 1972,
+    "肢": 2160,
+    "肤": 2708,
+    "肥": 1872,
+    "肩": 1516,
+    "肪": 2973,
+    "肯": 942,
+    "肱": 3819,
+    "育": 573,
+    "肴": 2147,
+    "肺": 1331,
+    "肼": 3161,
+    "肽": 2775,
+    "肾": 2604,
+    "肿": 2127,
+    "胀": 2167,
+    "胁": 847,
+    "胃": 2841,
+    "胄": 3205,
+    "胆": 2011,
+    "背": 780,
+    "胍": 3845,
+    "胎": 1547,
+    "胖": 3378,
+    "胚": 3460,
+    "胛": 4361,
+    "胜": 659,
+    "胞": 803,
+    "胡": 968,
+    "胤": 3075,
+    "胥": 4463,
+    "胪": 3925,
+    "胭": 3834,
+    "胯": 4039,
+    "胰": 3265,
+    "胱": 3259,
+    "胳": 4400,
+    "胶": 1572,
+    "胸": 2448,
+    "胺": 2235,
+    "胼": 4468,
+    "能": 26,
+    "脂": 2489,
+    "脆": 3707,
+    "脉": 966,
+    "脊": 2502,
+    "脏": 1159,
+    "脐": 3469,
+    "脑": 774,
+    "脓": 3286,
+    "脖": 3529,
+    "脚": 1091,
+    "脱": 1376,
+    "脲": 4151,
+    "脸": 1668,
+    "脾": 4355,
+    "腈": 3796,
+    "腊": 837,
+    "腐": 1710,
+    "腓": 2033,
+    "腔": 2486,
+    "腕": 3122,
+    "腥": 3303,
+    "腧": 3335,
+    "腩": 4283,
+    "腭": 2860,
+    "腮": 3679,
+    "腰": 1528,
+    "腱": 3958,
+    "腹": 1759,
+    "腺": 1860,
+    "腻": 3835,
+    "腾": 2601,
+    "腿": 1761,
+    "膀": 1505,
+    "膈": 4038,
+    "膊": 4401,
+    "膏": 4447,
+    "膑": 3897,
+    "膛": 4147,
+    "膜": 2475,
+    "膝": 2012,
+    "膦": 3595,
+    "膨": 2166,
+    "膳": 3297,
+    "膺": 3783,
+    "膻": 4026,
+    "臀": 3253,
+    "臂": 2222,
+    "臣": 1871,
+    "臧": 3053,
+    "自": 76,
+    "臭": 1585,
+    "臯": 4422,
+    "至": 219,
+    "致": 368,
+    "臻": 4234,
+    "臼": 4418,
+    "舄": 3278,
+    "舅": 3210,
+    "舆": 3398,
+    "舌": 2599,
+    "舍": 1247,
+    "舒": 1129,
+    "舘": 4219,
+    "舜": 2875,
+    "舞": 1148,
+    "舟": 2749,
+    "航": 264,
+    "舫": 4004,
+    "般": 919,
+    "舰": 711,
+    "舱": 755,
+    "舵": 4068,
+    "舶": 3111,
+    "船": 393,
+    "艇": 645,
+    "艘": 1040,
+    "艮": 3926,
+    "良": 1248,
+    "艰": 2296,
+    "色": 609,
+    "艳": 2207,
+    "艷": 4007,
+    "艺": 787,
+    "艾": 931,
+    "节": 242,
+    "芃": 3702,
+    "芋": 2596,
+    "芎": 3792,
+    "芒": 2413,
+    "芗": 4232,
+    "芙": 2878,
+    "芝": 1870,
+    "芥": 2634,
+    "芦": 2552,
+    "芩": 2900,
+    "芪": 2862,
+    "芬": 1024,
+    "芭": 2792,
+    "芮": 3364,
+    "芯": 1573,
+    "芰": 4313,
+    "花": 635,
+    "芳": 2510,
+    "芷": 4369,
+    "芸": 2924,
+    "芹": 2589,
+    "芽": 2769,
+    "苁": 3799,
+    "苇": 3784,
+    "苈": 3074,
+    "苋": 2852,
+    "苍": 2701,
+    "苎": 3185,
+    "苏": 592,
+    "苑": 2690,
+    "苓": 4102,
+    "苔": 2426,
+    "苗": 1561,
+    "苛": 4131,
+    "苞": 2632,
+    "苟": 3732,
+    "苡": 4278,
+    "苣": 2422,
+    "若": 1918,
+    "苦": 1279,
+    "苯": 2659,
+    "英": 172,
+    "苳": 3859,
+    "苴": 3451,
+    "苷": 3594,
+    "苹": 1399,
+    "茁": 4010,
+    "茂": 2429,
+    "范": 947,
+    "茄": 2678,
+    "茅": 2472,
+    "茉": 3697,
+    "茎": 2481,
+    "茛": 2479,
+    "茜": 1874,
+    "茧": 3907,
+    "茨": 1084,
+    "茫": 3785,
+    "茱": 2932,
+    "茵": 2183,
+    "茶": 1829,
+    "茹": 2720,
+    "荀": 2996,
+    "荁": 4034,
+    "荃": 3761,
+    "荆": 2505,
+    "草": 1141,
+    "荐": 1497,
+    "荒": 1559,
+    "荔": 3260,
+    "荚": 2760,
+    "荛": 3663,
+    "荠": 2697,
+    "荡": 1496,
+    "荣": 1270,
+    "荥": 3719,
+    "荦": 3868,
+    "荧": 4453,
+    "荨": 2681,
+    "荪": 4201,
+    "荫": 2960,
+    "药": 1398,
+    "荷": 1151,
+    "荸": 3154,
+    "荼": 3936,
+    "荽": 3700,
+    "莅": 3494,
+    "莆": 2883,
+    "莉": 2287,
+    "莎": 1448,
+    "莒": 3505,
+    "莓": 2830,
+    "莘": 3180,
+    "莞": 4410,
+    "莨": 2214,
+    "莩": 3667,
+    "莪": 2941,
+    "莫": 986,
+    "莱": 840,
+    "莲": 2407,
+    "莴": 3346,
+    "获": 298,
+    "莹": 2799,
+    "莺": 3304,
+    "莼": 3427,
+    "莽": 3514,
+    "菀": 2824,
+    "菁": 3171,
+    "菅": 3090,
+    "菇": 3281,
+    "菉": 4035,
+    "菊": 2178,
+    "菌": 905,
+    "菏": 3360,
+    "菜": 1072,
+    "菝": 3167,
+    "菠": 3574,
+    "菩": 2751,
+    "菪": 2215,
+    "菱": 2814,
+    "菲": 1160,
+    "萁": 4231,
+    "萃": 4052,
+    "萄": 1456,
+    "萌": 3057,
+    "萍": 2947,
+    "萎": 3472,
+    "萘": 4308,
+    "萝": 2712,
+    "萤": 2474,
+    "营": 576,
+    "萦": 2002,
+    "萧": 2440,
+    "萨": 432,
+    "萱": 3403,
+    "萸": 3138,
+    "萼": 2497,
+    "落": 546,
+    "葆": 3112,
+    "葎": 3419,
+    "著": 677,
+    "葛": 2462,
+    "葜": 3168,
+    "葡": 1455,
+    "董": 2376,
+    "葫": 2912,
+    "葬": 1251,
+    "葱": 1645,
+    "葳": 4249,
+    "葵": 2507,
+    "葶": 3073,
+    "蒂": 695,
+    "蒋": 2525,
+    "蒙": 877,
+    "蒜": 2772,
+    "蒟": 3971,
+    "蒲": 2461,
+    "蒴": 3331,
+    "蒸": 1987,
+    "蒺": 4241,
+    "蒿": 2535,
+    "蓄": 2155,
+    "蓉": 2913,
+    "蓍": 4319,
+    "蓑": 4328,
+    "蓝": 2381,
+    "蓟": 2828,
+    "蓣": 3658,
+    "蓬": 2580,
+    "蓼": 2673,
+    "蔑": 4124,
+    "蔓": 2340,
+    "蔗": 2091,
+    "蔚": 3081,
+    "蔡": 2317,
+    "蔬": 2864,
+    "蔵": 4253,
+    "蔷": 2517,
+    "蔺": 3232,
+    "蔻": 3486,
+    "蔽": 3095,
+    "蕃": 2915,
+    "蕈": 2756,
+    "蕉": 2639,
+    "蕊": 2761,
+    "蕗": 3713,
+    "蕨": 2351,
+    "蕲": 3683,
+    "蕴": 2853,
+    "蕾": 2329,
+    "薄": 1328,
+    "薇": 2482,
+    "薖": 4209,
+    "薛": 2672,
+    "薨": 4135,
+    "薪": 2201,
+    "薮": 3775,
+    "薯": 2627,
+    "薷": 4291,
+    "薹": 2614,
+    "藉": 1901,
+    "藏": 969,
+    "藓": 2379,
+    "藔": 4441,
+    "藜": 2901,
+    "藤": 1555,
+    "藨": 2827,
+    "藩": 2434,
+    "藳": 4206,
+    "藻": 1408,
+    "藿": 2910,
+    "蘑": 3608,
+    "蘸": 4061,
+    "虎": 1495,
+    "虏": 2260,
+    "虐": 1308,
+    "虑": 956,
+    "虔": 2248,
+    "虚": 1499,
+    "虞": 2972,
+    "虫": 971,
+    "虬": 3912,
+    "虱": 2662,
+    "虹": 1615,
+    "虻": 3810,
+    "虽": 651,
+    "虾": 2397,
+    "蚀": 1968,
+    "蚁": 1413,
+    "蚂": 2305,
+    "蚊": 1313,
+    "蚌": 2765,
+    "蚓": 2568,
+    "蚕": 2710,
+    "蚜": 2548,
+    "蚣": 3456,
+    "蚤": 2439,
+    "蚪": 4042,
+    "蚬": 2797,
+    "蚶": 3327,
+    "蚺": 3898,
+    "蛄": 2591,
+    "蛇": 1838,
+    "蛉": 4353,
+    "蛊": 4181,
+    "蛋": 1601,
+    "蛎": 3603,
+    "蛏": 3465,
+    "蛙": 2629,
+    "蛛": 2353,
+    "蛟": 3558,
+    "蛤": 2478,
+    "蛭": 2754,
+    "蛮": 2851,
+    "蛱": 2640,
+    "蛲": 4260,
+    "蛳": 3443,
+    "蛸": 3610,
+    "蛹": 3348,
+    "蛾": 2430,
+    "蜀": 2671,
+    "蜂": 2103,
+    "蜃": 1836,
+    "蜈": 3455,
+    "蜊": 3493,
+    "蜍": 2990,
+    "蜑": 4421,
+    "蜒": 1970,
+    "蜓": 3573,
+    "蜕": 3226,
+    "蜗": 2150,
+    "蜘": 2963,
+    "蜜": 2514,
+    "蜡": 2582,
+    "蜥": 2527,
+    "蜱": 2473,
+    "蜴": 3492,
+    "蜻": 3572,
+    "蜿": 1969,
+    "蝇": 2847,
+    "蝉": 2655,
+    "蝌": 4041,
+    "蝎": 2233,
+    "蝙": 2939,
+    "蝠": 2506,
+    "蝮": 4329,
+    "蝴": 3047,
+    "蝶": 2371,
+    "蝽": 2244,
+    "蝾": 4046,
+    "螂": 3831,
+    "螃": 4179,
+    "螈": 3397,
+    "融": 1047,
+    "螟": 3959,
+    "螣": 4065,
+    "螨": 3462,
+    "螯": 2899,
+    "螺": 2006,
+    "蟀": 4405,
+    "蟆": 4373,
+    "蟋": 4404,
+    "蟑": 3830,
+    "蟒": 2992,
+    "蟥": 3997,
+    "蟹": 2355,
+    "蟾": 2826,
+    "蠋": 4265,
+    "蠕": 3559,
+    "蠢": 1976,
+    "蠹": 4172,
+    "血": 1028,
+    "衅": 3411,
+    "行": 32,
+    "衍": 2624,
+    "衔": 1492,
+    "街": 902,
+    "衙": 2985,
+    "衡": 1232,
+    "衢": 3786,
+    "衣": 1366,
+    "补": 1099,
+    "表": 104,
+    "衫": 4443,
+    "衬": 3762,
+    "衰": 2667,
+    "衷": 3022,
+    "袁": 2544,
+    "袋": 1688,
+    "袍": 3026,
+    "袒": 4492,
+    "袓": 3975,
+    "袖": 1814,
+    "袜": 2334,
+    "被": 70,
+    "袭": 827,
+    "裁": 1032,
+    "裂": 924,
+    "装": 534,
+    "裔": 1593,
+    "裕": 1861,
+    "裘": 3821,
+    "裙": 1586,
+    "裤": 2059,
+    "裬": 4259,
+    "裴": 2778,
+    "裸": 1956,
+    "裹": 1787,
+    "褐": 2622,
+    "褒": 3485,
+    "褓": 4107,
+    "褚": 3513,
+    "褧": 4185,
+    "褪": 4431,
+    "褶": 2843,
+    "襄": 2786,
+    "襟": 3430,
+    "西": 163,
+    "要": 44,
+    "覃": 3337,
+    "覆": 1214,
+    "见": 462,
+    "观": 431,
+    "规": 385,
+    "觅": 3243,
+    "视": 399,
+    "览": 1042,
+    "觉": 870,
+    "角": 714,
+    "觚": 3888,
+    "解": 343,
+    "触": 933,
+    "言": 358,
+    "詹": 1550,
+    "誉": 1424,
+    "誓": 2954,
+    "諡": 3963,
+    "諲": 3883,
+    "謇": 4346,
+    "警": 449,
+    "譬": 4085,
+    "讚": 3957,
+    "计": 275,
+    "订": 844,
+    "认": 236,
+    "讨": 1518,
+    "让": 292,
+    "讫": 4073,
+    "训": 882,
+    "议": 210,
+    "讯": 1352,
+    "记": 415,
+    "讲": 822,
+    "讳": 2807,
+    "讶": 3567,
+    "讷": 2411,
+    "许": 229,
+    "讹": 4118,
+    "论": 476,
+    "讼": 1349,
+    "讽": 1646,
+    "设": 347,
+    "访": 729,
+    "诀": 2129,
+    "证": 335,
+    "诃": 3143,
+    "评": 794,
+    "诅": 2084,
+    "识": 530,
+    "诈": 1953,
+    "诉": 562,
+    "诊": 1240,
+    "词": 736,
+    "诏": 2619,
+    "译": 1259,
+    "试": 483,
+    "诗": 1674,
+    "诙": 4062,
+    "诚": 1756,
+    "诛": 2983,
+    "话": 487,
+    "诞": 1995,
+    "诟": 3550,
+    "诠": 3778,
+    "诡": 3699,
+    "询": 1805,
+    "该": 91,
+    "详": 1442,
+    "诩": 2229,
+    "诫": 3645,
+    "诬": 4112,
+    "语": 129,
+    "误": 1461,
+    "诱": 2067,
+    "说": 187,
+    "诵": 3708,
+    "请": 574,
+    "诸": 1234,
+    "诹": 4084,
+    "诺": 616,
+    "读": 604,
+    "诽": 3508,
+    "课": 1253,
+    "谁": 2789,
+    "调": 552,
+    "谅": 3233,
+    "谈": 1369,
+    "谊": 2771,
+    "谋": 1481,
+    "谌": 4196,
+    "谍": 3059,
+    "谎": 1638,
+    "谏": 2959,
+    "谐": 1788,
+    "谒": 3665,
+    "谓": 2579,
+    "谕": 3275,
+    "谚": 3704,
+    "谛": 3321,
+    "谜": 3216,
+    "谟": 2747,
+    "谠": 3881,
+    "谢": 1385,
+    "谣": 2967,
+    "谤": 3256,
+    "谥": 2969,
+    "谦": 2534,
+    "谨": 1754,
+    "谪": 4423,
+    "谬": 3131,
+    "谭": 2763,
+    "谯": 4023,
+    "谱": 2484,
+    "谴": 2952,
+    "谶": 3512,
+    "谷": 873,
+    "豁": 2156,
+    "豆": 2171,
+    "豌": 3237,
+    "豚": 3045,
+    "象": 599,
+    "豪": 1664,
+    "豫": 2844,
+    "豹": 1201,
+    "貂": 3551,
+    "貌": 1536,
+    "貘": 4314,
+    "贝": 910,
+    "贞": 2416,
+    "负": 885,
+    "贡": 1770,
+    "财": 1105,
+    "责": 786,
+    "贤": 2555,
+    "败": 722,
+    "账": 3749,
+    "货": 957,
+    "质": 427,
+    "贩": 2795,
+    "贪": 1662,
+    "贫": 1818,
+    "贬": 2223,
+    "购": 798,
+    "贮": 4103,
+    "贯": 1530,
+    "贰": 3481,
+    "贴": 1589,
+    "贵": 1196,
+    "贷": 2157,
+    "贸": 1294,
+    "费": 508,
+    "贺": 1673,
+    "贻": 3808,
+    "贼": 2303,
+    "贾": 1702,
+    "贿": 2285,
+    "赁": 2241,
+    "赂": 3509,
+    "赃": 3705,
+    "资": 621,
+    "赈": 3313,
+    "赉": 3415,
+    "赋": 2570,
+    "赌": 1226,
+    "赎": 3780,
+    "赏": 1362,
+    "赐": 2532,
+    "赔": 2098,
+    "赖": 1224,
+    "赘": 4200,
+    "赚": 2669,
+    "赛": 177,
+    "赞": 1202,
+    "赟": 3511,
+    "赠": 1867,
+    "赡": 4235,
+    "赢": 1027,
+    "赣": 2933,
+    "赤": 1684,
+    "赦": 3010,
+    "赫": 745,
+    "赭": 3476,
+    "走": 624,
+    "赴": 2070,
+    "赵": 2378,
+    "赶": 2069,
+    "起": 180,
+    "趁": 1884,
+    "超": 579,
+    "越": 445,
+    "趋": 2707,
+    "趟": 2029,
+    "趣": 911,
+    "足": 375,
+    "趴": 4477,
+    "趾": 1985,
+    "跃": 1437,
+    "跆": 3067,
+    "跋": 2302,
+    "跌": 1701,
+    "跑": 699,
+    "跖": 3436,
+    "距": 691,
+    "跟": 1269,
+    "跨": 991,
+    "跪": 4436,
+    "路": 348,
+    "跳": 1735,
+    "践": 1703,
+    "跻": 4044,
+    "踊": 3960,
+    "踏": 2752,
+    "踢": 1876,
+    "踩": 3909,
+    "踪": 1525,
+    "踵": 2311,
+    "蹄": 2327,
+    "蹈": 1851,
+    "蹴": 4092,
+    "蹶": 3892,
+    "身": 282,
+    "躯": 3993,
+    "躲": 2216,
+    "躺": 4437,
+    "车": 97,
+    "轧": 4339,
+    "轨": 1717,
+    "轩": 2943,
+    "转": 501,
+    "轭": 4307,
+    "轮": 836,
+    "软": 1181,
+    "轰": 3037,
+    "轲": 4070,
+    "轴": 1642,
+    "轶": 3882,
+    "轸": 3510,
+    "轻": 570,
+    "载": 976,
+    "轿": 1809,
+    "较": 494,
+    "辅": 2158,
+    "辆": 1056,
+    "辇": 3470,
+    "辈": 1879,
+    "辉": 2541,
+    "辍": 3770,
+    "辐": 1382,
+    "辑": 1335,
+    "输": 884,
+    "辕": 3939,
+    "辖": 2349,
+    "辗": 3202,
+    "辙": 3377,
+    "辛": 1457,
+    "辜": 3745,
+    "辞": 1694,
+    "辟": 2887,
+    "辣": 3215,
+    "辨": 1651,
+    "辩": 2988,
+    "辰": 2460,
+    "辱": 3600,
+    "边": 514,
+    "込": 4166,
+    "辽": 2392,
+    "达": 171,
+    "迁": 1686,
+    "迄": 2123,
+    "迅": 1146,
+    "过": 61,
+    "迈": 1822,
+    "迎": 975,
+    "运": 257,
+    "近": 352,
+    "返": 880,
+    "还": 196,
+    "这": 12,
+    "进": 113,
+    "远": 448,
+    "违": 1075,
+    "连": 644,
+    "迟": 1377,
+    "迦": 2569,
+    "迩": 2056,
+    "迪": 764,
+    "迫": 865,
+    "迭": 3421,
+    "述": 929,
+    "迷": 2344,
+    "迹": 1037,
+    "追": 715,
+    "退": 955,
+    "送": 653,
+    "适": 507,
+    "逃": 1113,
+    "逆": 1778,
+    "选": 251,
+    "逊": 1106,
+    "逍": 2893,
+    "透": 1060,
+    "逐": 1011,
+    "递": 2572,
+    "途": 1057,
+    "逗": 1726,
+    "通": 83,
+    "逛": 2200,
+    "逝": 1549,
+    "速": 321,
+    "造": 295,
+    "逢": 3369,
+    "逮": 1504,
+    "逸": 2689,
+    "逻": 1705,
+    "逼": 2100,
+    "遁": 3771,
+    "遂": 2078,
+    "遇": 776,
+    "遍": 1678,
+    "遏": 2249,
+    "遐": 2055,
+    "遑": 4169,
+    "道": 140,
+    "遗": 750,
+    "遣": 1916,
+    "遥": 1281,
+    "遭": 1068,
+    "遮": 1799,
+    "遵": 2312,
+    "避": 1008,
+    "邀": 1908,
+    "邑": 2610,
+    "邓": 1707,
+    "邕": 3428,
+    "邢": 2811,
+    "那": 230,
+    "邦": 1018,
+    "邨": 2905,
+    "邪": 2759,
+    "邬": 3696,
+    "邮": 841,
+    "邯": 3184,
+    "邰": 4294,
+    "邱": 2703,
+    "邳": 3586,
+    "邵": 2628,
+    "邸": 1937,
+    "邹": 3657,
+    "邺": 3640,
+    "邻": 973,
+    "邾": 3916,
+    "郁": 1597,
+    "郃": 3296,
+    "郊": 2313,
+    "郎": 2368,
+    "郑": 2380,
+    "郓": 3417,
+    "郝": 3654,
+    "郡": 2350,
+    "郤": 4389,
+    "郦": 3621,
+    "部": 77,
+    "郫": 3903,
+    "郭": 2404,
+    "郯": 3800,
+    "郴": 3880,
+    "郸": 3370,
+    "都": 79,
+    "鄂": 2800,
+    "鄙": 4040,
+    "鄞": 3385,
+    "鄣": 3527,
+    "鄱": 3815,
+    "酃": 4017,
+    "酉": 2781,
+    "酋": 1719,
+    "酌": 3605,
+    "配": 1169,
+    "酐": 3856,
+    "酒": 447,
+    "酚": 3115,
+    "酢": 4277,
+    "酥": 3043,
+    "酪": 3846,
+    "酬": 2945,
+    "酮": 2806,
+    "酯": 2618,
+    "酰": 2731,
+    "酱": 2242,
+    "酵": 2993,
+    "酶": 2495,
+    "酷": 2920,
+    "酸": 1489,
+    "酿": 2837,
+    "醇": 2281,
+    "醉": 2052,
+    "醋": 3809,
+    "醒": 1233,
+    "醚": 2894,
+    "醛": 2725,
+    "醮": 4271,
+    "醯": 4450,
+    "采": 961,
+    "釉": 3234,
+    "释": 1089,
+    "里": 47,
+    "重": 134,
+    "野": 467,
+    "量": 227,
+    "金": 263,
+    "釜": 2953,
+    "鉴": 1293,
+    "銮": 3186,
+    "鋆": 4009,
+    "鋐": 4099,
+    "鍊": 3530,
+    "鎏": 4368,
+    "鏊": 3644,
+    "鑫": 3144,
+    "针": 963,
+    "钉": 1374,
+    "钊": 4150,
+    "钌": 3890,
+    "钓": 3394,
+    "钕": 3307,
+    "钗": 4348,
+    "钙": 2277,
+    "钛": 2342,
+    "钜": 3426,
+    "钝": 2742,
+    "钟": 686,
+    "钠": 2813,
+    "钡": 3736,
+    "钢": 1500,
+    "钥": 2320,
+    "钦": 2540,
+    "钧": 3085,
+    "钨": 4304,
+    "钩": 2550,
+    "钪": 4255,
+    "钫": 4359,
+    "钮": 3217,
+    "钯": 3230,
+    "钰": 3349,
+    "钱": 965,
+    "钲": 3914,
+    "钳": 1685,
+    "钴": 4199,
+    "钵": 3358,
+    "钹": 4243,
+    "钻": 2204,
+    "钼": 4473,
+    "钾": 3231,
+    "钿": 4184,
+    "铀": 2261,
+    "铁": 1291,
+    "铂": 3754,
+    "铃": 2045,
+    "铅": 1815,
+    "铆": 3795,
+    "铋": 3769,
+    "铎": 2980,
+    "铜": 2035,
+    "铝": 2687,
+    "铟": 3488,
+    "铠": 2868,
+    "铣": 4467,
+    "铨": 3114,
+    "铬": 3457,
+    "铭": 2520,
+    "铮": 4380,
+    "铯": 4305,
+    "铰": 3759,
+    "铲": 4096,
+    "铳": 3522,
+    "铵": 3801,
+    "银": 793,
+    "铸": 2609,
+    "铺": 1610,
+    "铼": 3979,
+    "铽": 4138,
+    "链": 2493,
+    "销": 1207,
+    "锁": 1923,
+    "锂": 3193,
+    "锅": 1792,
+    "锆": 3751,
+    "锈": 3068,
+    "锋": 1957,
+    "锌": 3564,
+    "锐": 2231,
+    "锑": 4298,
+    "错": 1209,
+    "锚": 3118,
+    "锡": 1229,
+    "锣": 4333,
+    "锤": 2846,
+    "锥": 2543,
+    "锦": 1252,
+    "键": 1334,
+    "锯": 2798,
+    "锰": 3743,
+    "锶": 4325,
+    "锺": 3646,
+    "锻": 1618,
+    "镀": 3181,
+    "镁": 3287,
+    "镂": 4160,
+    "镇": 864,
+    "镉": 4015,
+    "镊": 3352,
+    "镍": 4014,
+    "镎": 4048,
+    "镐": 3777,
+    "镑": 1683,
+    "镒": 3528,
+    "镓": 4186,
+    "镖": 1361,
+    "镗": 4256,
+    "镛": 3533,
+    "镜": 465,
+    "镠": 4187,
+    "镥": 3849,
+    "镧": 3126,
+    "镫": 1831,
+    "镰": 2911,
+    "镳": 2198,
+    "镶": 2682,
+    "长": 178,
+    "閒": 3662,
+    "闍": 3414,
+    "门": 471,
+    "闪": 1578,
+    "闫": 3627,
+    "闭": 835,
+    "问": 306,
+    "闯": 3221,
+    "闰": 4013,
+    "闱": 4484,
+    "闲": 1342,
+    "闳": 3721,
+    "间": 103,
+    "闵": 2942,
+    "闸": 2646,
+    "闹": 1416,
+    "闻": 595,
+    "闼": 4352,
+    "闽": 2688,
+    "闾": 3591,
+    "阀": 1605,
+    "阁": 2419,
+    "阅": 861,
+    "阇": 3677,
+    "阈": 4279,
+    "阉": 4334,
+    "阎": 3040,
+    "阏": 4375,
+    "阐": 2997,
+    "阔": 2092,
+    "阕": 4069,
+    "阖": 4049,
+    "阙": 3383,
+    "阜": 2650,
+    "队": 170,
+    "阪": 1888,
+    "阮": 2398,
+    "阱": 1843,
+    "防": 580,
+    "阳": 603,
+    "阴": 2032,
+    "阵": 1287,
+    "阶": 1258,
+    "阻": 1186,
+    "阿": 271,
+    "陀": 2005,
+    "陁": 3425,
+    "陂": 3884,
+    "附": 600,
+    "际": 586,
+    "陆": 567,
+    "陇": 2951,
+    "陈": 2354,
+    "陋": 3401,
+    "陌": 1808,
+    "降": 547,
+    "限": 652,
+    "陕": 2399,
+    "陟": 4188,
+    "陡": 1026,
+    "院": 435,
+    "除": 629,
+    "陨": 1914,
+    "险": 566,
+    "陪": 1584,
+    "陲": 3709,
+    "陵": 1051,
+    "陶": 1687,
+    "陷": 1035,
+    "隅": 3127,
+    "隆": 1059,
+    "隈": 4149,
+    "隋": 2557,
+    "隍": 3007,
+    "随": 357,
+    "隐": 1102,
+    "隔": 1273,
+    "隗": 3877,
+    "隘": 4438,
+    "隙": 1592,
+    "障": 878,
+    "隧": 1924,
+    "隶": 2131,
+    "隼": 3084,
+    "难": 458,
+    "雀": 2149,
+    "雁": 3781,
+    "雄": 1847,
+    "雅": 1081,
+    "集": 650,
+    "雇": 1658,
+    "雉": 3336,
+    "雌": 2288,
+    "雍": 2487,
+    "雏": 3108,
+    "雑": 4115,
+    "雕": 1905,
+    "雨": 778,
+    "雪": 372,
+    "雯": 3367,
+    "雳": 4230,
+    "零": 2300,
+    "雷": 656,
+    "雾": 2603,
+    "需": 232,
+    "霄": 4310,
+    "霆": 3655,
+    "震": 760,
+    "霈": 3863,
+    "霉": 2716,
+    "霍": 1147,
+    "霓": 4370,
+    "霖": 3201,
+    "霜": 3390,
+    "霞": 2745,
+    "露": 710,
+    "霸": 2513,
+    "霹": 4229,
+    "青": 1737,
+    "靖": 2361,
+    "静": 1340,
+    "靛": 3357,
+    "非": 202,
+    "靠": 908,
+    "靡": 4420,
+    "面": 143,
+    "革": 1161,
+    "靳": 3966,
+    "靴": 1546,
+    "靶": 3772,
+    "靼": 3325,
+    "鞅": 3576,
+    "鞋": 848,
+    "鞍": 1712,
+    "鞑": 3324,
+    "鞘": 2663,
+    "鞣": 4452,
+    "鞭": 1894,
+    "韦": 1162,
+    "韧": 3203,
+    "韩": 1297,
+    "韬": 3376,
+    "韭": 4123,
+    "音": 518,
+    "韵": 2206,
+    "韶": 3192,
+    "頴": 4213,
+    "页": 1563,
+    "顶": 1043,
+    "顷": 2165,
+    "项": 355,
+    "顺": 1239,
+    "须": 538,
+    "顽": 4385,
+    "顾": 1205,
+    "顿": 728,
+    "颁": 1403,
+    "颂": 3149,
+    "预": 387,
+    "颅": 2323,
+    "领": 400,
+    "颇": 1837,
+    "颈": 2605,
+    "颉": 3123,
+    "颊": 3282,
+    "颌": 3069,
+    "颍": 3639,
+    "颐": 3501,
+    "频": 829,
+    "颓": 3454,
+    "颖": 2613,
+    "颗": 1438,
+    "题": 360,
+    "颜": 1581,
+    "额": 821,
+    "颞": 4267,
+    "颠": 1853,
+    "颤": 3140,
+    "风": 284,
+    "飒": 4109,
+    "飓": 1118,
+    "飘": 2053,
+    "飙": 4481,
+    "飞": 329,
+    "食": 543,
+    "餐": 815,
+    "餵": 4264,
+    "饥": 3314,
+    "饪": 1634,
+    "饭": 1502,
+    "饮": 777,
+    "饯": 4301,
+    "饰": 1153,
+    "饱": 1798,
+    "饲": 1899,
+    "饴": 4448,
+    "饵": 4479,
+    "饶": 2914,
+    "饷": 3459,
+    "饼": 2788,
+    "饿": 2998,
+    "馀": 3857,
+    "馅": 3477,
+    "馆": 713,
+    "馈": 2144,
+    "馊": 2080,
+    "馔": 4372,
+    "首": 474,
+    "香": 1166,
+    "馥": 4471,
+    "馨": 3623,
+    "馯": 4248,
+    "马": 179,
+    "驭": 3923,
+    "驯": 1796,
+    "驰": 1941,
+    "驱": 1036,
+    "驳": 1930,
+    "驴": 3160,
+    "驶": 751,
+    "驷": 4374,
+    "驸": 3482,
+    "驹": 2794,
+    "驻": 1748,
+    "驼": 1669,
+    "驾": 766,
+    "驿": 3196,
+    "骁": 3949,
+    "骂": 3919,
+    "骄": 3768,
+    "骆": 3499,
+    "骈": 4246,
+    "验": 569,
+    "骏": 2944,
+    "骑": 738,
+    "骗": 1347,
+    "骘": 4226,
+    "骚": 2202,
+    "骠": 3741,
+    "骤": 2848,
+    "骥": 3951,
+    "骨": 1276,
+    "骰": 3930,
+    "骷": 3895,
+    "骸": 2309,
+    "骼": 3097,
+    "髅": 3896,
+    "髎": 4095,
+    "髓": 3204,
+    "高": 209,
+    "鬃": 4340,
+    "鬣": 2986,
+    "鬼": 2519,
+    "魁": 2136,
+    "魂": 2680,
+    "魄": 3653,
+    "魅": 1877,
+    "魇": 3985,
+    "魏": 2367,
+    "魔": 2476,
+    "鮎": 3879,
+    "鱼": 1190,
+    "鱿": 3739,
+    "鲀": 4272,
+    "鲁": 619,
+    "鲃": 3266,
+    "鲈": 2766,
+    "鲍": 1803,
+    "鲎": 4148,
+    "鲑": 3395,
+    "鲜": 1490,
+    "鲡": 3236,
+    "鲢": 3701,
+    "鲣": 3219,
+    "鲤": 2615,
+    "鲨": 1830,
+    "鲫": 3155,
+    "鲱": 3652,
+    "鲳": 3453,
+    "鲷": 2636,
+    "鲸": 2744,
+    "鲹": 3188,
+    "鲻": 3483,
+    "鳃": 3862,
+    "鳄": 2879,
+    "鳅": 2480,
+    "鳌": 3612,
+    "鳍": 2423,
+    "鳔": 3418,
+    "鳕": 2882,
+    "鳖": 3049,
+    "鳗": 2664,
+    "鳚": 3934,
+    "鳞": 2384,
+    "鳟": 4446,
+    "鸟": 977,
+    "鸠": 3091,
+    "鸡": 1611,
+    "鸣": 2598,
+    "鸥": 3163,
+    "鸦": 1365,
+    "鸩": 4247,
+    "鸫": 2020,
+    "鸭": 1795,
+    "鸮": 3714,
+    "鸯": 4345,
+    "鸲": 4043,
+    "鸵": 4055,
+    "鸻": 3099,
+    "鸽": 3247,
+    "鸾": 3389,
+    "鸿": 2612,
+    "鹂": 3908,
+    "鹃": 2374,
+    "鹅": 1742,
+    "鹉": 2117,
+    "鹋": 3525,
+    "鹌": 3854,
+    "鹏": 3029,
+    "鹑": 3498,
+    "鹘": 2908,
+    "鹛": 3947,
+    "鹤": 2490,
+    "鹦": 2116,
+    "鹩": 3526,
+    "鹪": 4204,
+    "鹬": 4403,
+    "鹭": 3052,
+    "鹰": 1532,
+    "鹳": 3332,
+    "鹿": 1137,
+    "麂": 4488,
+    "麋": 1766,
+    "麒": 3628,
+    "麓": 2891,
+    "麝": 3209,
+    "麟": 2466,
+    "麦": 789,
+    "麻": 1275,
+    "麾": 1939,
+    "黄": 1566,
+    "黎": 1311,
+    "黏": 3015,
+    "黑": 693,
+    "黔": 3001,
+    "默": 1260,
+    "黛": 2917,
+    "黜": 4424,
+    "黧": 3452,
+    "黯": 3487,
+    "黾": 3921,
+    "鼎": 1841,
+    "鼐": 4263,
+    "鼓": 1400,
+    "鼠": 1604,
+    "鼢": 4227,
+    "鼩": 2595,
+    "鼬": 3552,
+    "鼱": 2809,
+    "鼷": 4080,
+    "鼹": 3096,
+    "鼻": 1881,
+    "齐": 1115,
+    "齿": 1533,
+    "龄": 1306,
+    "龈": 3293,
+    "龙": 625,
+    "龚": 4047,
+    "龛": 4445,
+    "龟": 2442,
+    "％": 1711,
+    "：": 1900,
+    "；": 2106
+  },
+  "cmo-script_khmer": {
+    "-": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 10,
+    "ក": 18,
+    "ខ": 39,
+    "គ": 29,
+    "ង": 13,
+    "ច": 40,
+    "ឆ": 35,
+    "ជ": 31,
+    "ញ": 33,
+    "ដ": 41,
+    "ត": 15,
+    "ថ": 47,
+    "ទ": 24,
+    "ន": 5,
+    "ប": 32,
+    "ផ": 42,
+    "ព": 30,
+    "ម": 8,
+    "យ": 22,
+    "រ": 9,
+    "ល": 16,
+    "វ": 26,
+    "ឞ": 17,
+    "ស": 7,
+    "ហ": 37,
+    "អ": 11,
+    "ា": 4,
+    "ិ": 46,
+    "ី": 20,
+    "ឹ": 36,
+    "ឺ": 45,
+    "ុ": 34,
+    "ូ": 27,
+    "ើ": 21,
+    "ៀ": 51,
+    "េ": 28,
+    "ែ": 25,
+    "ៃ": 23,
+    "ោ": 19,
+    "ៅ": 44,
+    "ៈ": 38,
+    "់": 6,
+    "្": 12,
+    "៝": 14,
+    "០": 43,
+    "១": 48,
+    "២": 50,
+    "៣": 54,
+    "៤": 52,
+    "៥": 53,
+    "៦": 55,
+    "៧": 57,
+    "៨": 58,
+    "៩": 56
+  },
+  "cmo-script_latin": {
+    "'": 35,
+    "-": 42,
+    "0": 39,
+    "1": 41,
+    "2": 43,
+    "3": 46,
+    "4": 44,
+    "5": 45,
+    "6": 47,
+    "7": 48,
+    "8": 50,
+    "9": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 32,
+    "d": 22,
+    "e": 26,
+    "g": 8,
+    "h": 7,
+    "i": 14,
+    "j": 25,
+    "k": 13,
+    "l": 18,
+    "m": 9,
+    "n": 5,
+    "o": 27,
+    "p": 19,
+    "r": 10,
+    "s": 28,
+    "t": 15,
+    "u": 11,
+    "v": 37,
+    "y": 17,
+    "|": 4,
+    "â": 12,
+    "ê": 24,
+    "ô": 23,
+    "ă": 16,
+    "đ": 31,
+    "ĕ": 33,
+    "ĭ": 38,
+    "ŏ": 29,
+    "ŭ": 30,
+    "ƀ": 34,
+    "ơ": 21,
+    "ư": 40,
+    "̆": 36
+  },
+  "cmr": {
+    "'": 28,
+    "-": 29,
+    "0": 32,
+    "1": 33,
+    "2": 34,
+    "3": 37,
+    "4": 35,
+    "6": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "d": 14,
+    "e": 8,
+    "f": 31,
+    "g": 21,
+    "h": 9,
+    "i": 7,
+    "j": 16,
+    "k": 10,
+    "l": 19,
+    "m": 11,
+    "n": 6,
+    "o": 13,
+    "p": 23,
+    "q": 12,
+    "r": 25,
+    "s": 22,
+    "t": 17,
+    "u": 18,
+    "v": 27,
+    "w": 15,
+    "x": 24,
+    "y": 20,
+    "z": 30,
+    "|": 4
+  },
+  "cnh": {
+    "'": 28,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 16,
+    "d": 22,
+    "e": 15,
+    "f": 27,
+    "g": 14,
+    "h": 7,
+    "i": 8,
+    "j": 29,
+    "k": 9,
+    "l": 11,
+    "m": 12,
+    "n": 6,
+    "o": 19,
+    "p": 20,
+    "r": 17,
+    "s": 21,
+    "t": 13,
+    "u": 10,
+    "v": 25,
+    "w": 18,
+    "y": 30,
+    "z": 23,
+    "|": 4,
+    "ṭ": 26,
+    "—": 31
+  },
+  "cni": {
+    "'": 24,
+    "-": 22,
+    "0": 23,
+    "1": 25,
+    "2": 26,
+    "3": 32,
+    "4": 28,
+    "5": 30,
+    "6": 29,
+    "7": 27,
+    "8": 31,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 13,
+    "e": 7,
+    "h": 19,
+    "i": 5,
+    "j": 18,
+    "m": 17,
+    "n": 9,
+    "o": 11,
+    "p": 14,
+    "q": 15,
+    "r": 10,
+    "s": 12,
+    "t": 8,
+    "u": 16,
+    "v": 20,
+    "y": 21,
+    "|": 6
+  },
+  "cnl": {
+    "'": 7,
+    "1": 25,
+    "2": 6,
+    "3": 4,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 38,
+    "c": 18,
+    "d": 22,
+    "e": 9,
+    "f": 45,
+    "g": 35,
+    "h": 39,
+    "i": 10,
+    "j": 13,
+    "k": 47,
+    "l": 20,
+    "m": 21,
+    "n": 11,
+    "o": 15,
+    "p": 41,
+    "q": 23,
+    "r": 19,
+    "s": 16,
+    "t": 24,
+    "u": 12,
+    "v": 46,
+    "x": 26,
+    "y": 42,
+    "z": 44,
+    "|": 5,
+    "á": 32,
+    "é": 28,
+    "í": 33,
+    "ñ": 34,
+    "ó": 37,
+    "ø": 27,
+    "ú": 30,
+    "ü": 43,
+    "ŋ": 40,
+    "ǿ": 36,
+    "ɨ": 14,
+    "ʉ": 31,
+    "́": 29,
+    "̱": 17
+  },
+  "cnt": {
+    "'": 7,
+    "-": 40,
+    "1": 21,
+    "2": 5,
+    "3": 14,
+    "4": 12,
+    "5": 13,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 33,
+    "c": 18,
+    "d": 24,
+    "e": 16,
+    "f": 41,
+    "g": 15,
+    "h": 35,
+    "i": 8,
+    "j": 10,
+    "k": 45,
+    "l": 22,
+    "m": 26,
+    "n": 9,
+    "o": 23,
+    "p": 38,
+    "q": 29,
+    "r": 34,
+    "s": 17,
+    "t": 27,
+    "u": 11,
+    "v": 42,
+    "x": 46,
+    "y": 44,
+    "z": 43,
+    "|": 4,
+    "á": 19,
+    "é": 31,
+    "ë": 25,
+    "í": 37,
+    "ñ": 47,
+    "ó": 30,
+    "ú": 36,
+    "ü": 39,
+    "ŋ": 32,
+    "ɨ": 20,
+    "́": 28
+  },
+  "coe": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 35,
+    "c": 10,
+    "d": 26,
+    "e": 7,
+    "f": 38,
+    "g": 36,
+    "h": 12,
+    "i": 9,
+    "j": 15,
+    "k": 19,
+    "l": 31,
+    "m": 17,
+    "n": 13,
+    "o": 11,
+    "p": 18,
+    "q": 40,
+    "r": 14,
+    "s": 16,
+    "t": 21,
+    "u": 20,
+    "v": 24,
+    "x": 44,
+    "y": 43,
+    "z": 42,
+    "|": 5,
+    "á": 39,
+    "ã": 22,
+    "é": 37,
+    "í": 34,
+    "ñ": 23,
+    "ó": 41,
+    "õ": 32,
+    "ú": 33,
+    "ĩ": 29,
+    "ũ": 25,
+    "ʉ": 6,
+    "̃": 27,
+    "ẽ": 28,
+    "—": 30
+  },
+  "cof": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 10,
+    "d": 22,
+    "e": 8,
+    "f": 25,
+    "g": 32,
+    "h": 18,
+    "i": 6,
+    "j": 15,
+    "l": 13,
+    "m": 17,
+    "n": 7,
+    "o": 12,
+    "p": 21,
+    "q": 20,
+    "r": 16,
+    "s": 14,
+    "t": 11,
+    "u": 9,
+    "v": 33,
+    "w": 24,
+    "x": 35,
+    "y": 19,
+    "z": 34,
+    "|": 5,
+    "á": 26,
+    "é": 28,
+    "í": 27,
+    "ñ": 36,
+    "ó": 31,
+    "ú": 30,
+    "—": 29
+  },
+  "cok": {
+    "'": 7,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 16,
+    "d": 30,
+    "e": 11,
+    "f": 36,
+    "g": 35,
+    "h": 20,
+    "i": 13,
+    "j": 6,
+    "k": 37,
+    "l": 33,
+    "m": 15,
+    "n": 14,
+    "o": 22,
+    "p": 21,
+    "q": 31,
+    "r": 18,
+    "s": 19,
+    "t": 8,
+    "u": 10,
+    "v": 28,
+    "x": 25,
+    "y": 12,
+    "z": 26,
+    "|": 4,
+    "á": 17,
+    "é": 24,
+    "í": 27,
+    "ñ": 38,
+    "ó": 32,
+    "ú": 29,
+    "ɨ": 9,
+    "́": 23
+  },
+  "con": {
+    "'": 10,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 11,
+    "d": 22,
+    "e": 6,
+    "f": 21,
+    "g": 19,
+    "h": 18,
+    "i": 7,
+    "j": 16,
+    "k": 38,
+    "l": 29,
+    "m": 15,
+    "n": 8,
+    "o": 14,
+    "p": 20,
+    "q": 17,
+    "r": 27,
+    "s": 9,
+    "t": 12,
+    "u": 13,
+    "v": 26,
+    "x": 37,
+    "y": 23,
+    "z": 33,
+    "|": 5,
+    "á": 34,
+    "é": 31,
+    "í": 35,
+    "ñ": 24,
+    "ó": 36,
+    "ú": 28,
+    "ḿ": 39,
+    "—": 30
+  },
+  "cot": {
+    "'": 27,
+    "0": 24,
+    "1": 25,
+    "2": 29,
+    "3": 28,
+    "4": 31,
+    "5": 32,
+    "6": 33,
+    "7": 30,
+    "8": 35,
+    "9": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 37,
+    "c": 14,
+    "d": 41,
+    "e": 6,
+    "f": 42,
+    "g": 19,
+    "h": 20,
+    "i": 5,
+    "j": 15,
+    "l": 34,
+    "m": 17,
+    "n": 9,
+    "o": 11,
+    "p": 18,
+    "q": 12,
+    "r": 13,
+    "s": 16,
+    "t": 8,
+    "u": 10,
+    "v": 21,
+    "x": 44,
+    "y": 22,
+    "z": 38,
+    "|": 7,
+    "é": 40,
+    "í": 43,
+    "ñ": 26,
+    "ó": 39,
+    "—": 23
+  },
+  "cou": {
+    "'": 36,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 33,
+    "c": 29,
+    "d": 25,
+    "e": 11,
+    "f": 22,
+    "g": 27,
+    "h": 8,
+    "i": 9,
+    "j": 32,
+    "k": 10,
+    "l": 17,
+    "m": 20,
+    "n": 7,
+    "o": 13,
+    "p": 28,
+    "r": 16,
+    "s": 21,
+    "t": 15,
+    "u": 14,
+    "v": 18,
+    "w": 12,
+    "y": 19,
+    "|": 4,
+    "á": 38,
+    "ë": 6,
+    "ñ": 35,
+    "ŋ": 26,
+    "ƴ": 34,
+    "ɓ": 31,
+    "ɗ": 23,
+    "̃": 24,
+    "ỹ": 30,
+    "—": 39
+  },
+  "cpa": {
+    "1": 7,
+    "2": 6,
+    "3": 11,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 31,
+    "c": 17,
+    "d": 15,
+    "e": 25,
+    "f": 37,
+    "g": 14,
+    "h": 9,
+    "i": 8,
+    "j": 12,
+    "k": 43,
+    "l": 19,
+    "m": 20,
+    "n": 10,
+    "o": 21,
+    "p": 36,
+    "q": 27,
+    "r": 32,
+    "s": 13,
+    "t": 22,
+    "u": 16,
+    "v": 39,
+    "x": 42,
+    "y": 41,
+    "z": 38,
+    "|": 4,
+    "á": 23,
+    "é": 28,
+    "ë": 34,
+    "í": 29,
+    "ñ": 44,
+    "ó": 33,
+    "ø": 18,
+    "ú": 26,
+    "ü": 40,
+    "ŋ": 30,
+    "ǿ": 24,
+    "́": 35
+  },
+  "cpb": {
+    "'": 43,
+    "-": 24,
+    "0": 34,
+    "1": 36,
+    "2": 40,
+    "3": 48,
+    "4": 41,
+    "5": 42,
+    "6": 44,
+    "7": 39,
+    "8": 45,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 29,
+    "c": 22,
+    "d": 28,
+    "e": 9,
+    "f": 33,
+    "g": 23,
+    "h": 18,
+    "i": 5,
+    "j": 20,
+    "k": 11,
+    "l": 25,
+    "m": 15,
+    "n": 10,
+    "o": 12,
+    "p": 14,
+    "q": 38,
+    "r": 8,
+    "s": 13,
+    "t": 7,
+    "u": 26,
+    "v": 37,
+    "w": 16,
+    "x": 46,
+    "y": 17,
+    "z": 19,
+    "|": 6,
+    "á": 32,
+    "é": 30,
+    "í": 31,
+    "ñ": 21,
+    "ó": 35,
+    "ú": 27
+  },
+  "cpu": {
+    "-": 23,
+    "0": 33,
+    "1": 36,
+    "2": 39,
+    "3": 45,
+    "4": 40,
+    "5": 41,
+    "6": 42,
+    "7": 38,
+    "8": 43,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 22,
+    "d": 27,
+    "e": 9,
+    "f": 32,
+    "g": 29,
+    "h": 17,
+    "i": 5,
+    "j": 20,
+    "k": 11,
+    "l": 24,
+    "m": 15,
+    "n": 10,
+    "o": 12,
+    "p": 13,
+    "q": 37,
+    "r": 8,
+    "s": 14,
+    "t": 7,
+    "u": 25,
+    "v": 18,
+    "x": 44,
+    "y": 16,
+    "z": 19,
+    "|": 6,
+    "á": 35,
+    "é": 31,
+    "í": 30,
+    "ñ": 21,
+    "ó": 34,
+    "ú": 26
+  },
+  "crh": {
+    "-": 37,
+    "0": 38,
+    "1": 41,
+    "2": 39,
+    "3": 44,
+    "4": 42,
+    "5": 46,
+    "6": 45,
+    "7": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 19,
+    "в": 28,
+    "г": 20,
+    "д": 14,
+    "е": 7,
+    "ж": 32,
+    "з": 22,
+    "и": 8,
+    "й": 25,
+    "к": 13,
+    "л": 9,
+    "м": 15,
+    "н": 6,
+    "о": 18,
+    "п": 24,
+    "р": 10,
+    "с": 17,
+    "т": 16,
+    "у": 21,
+    "ф": 36,
+    "х": 33,
+    "ц": 40,
+    "ч": 29,
+    "ш": 23,
+    "ъ": 11,
+    "ы": 12,
+    "ь": 31,
+    "э": 27,
+    "ю": 26,
+    "я": 30,
+    "ё": 35,
+    "–": 34
+  },
+  "crk-script_latin": {
+    "'": 34,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 30,
+    "c": 22,
+    "d": 29,
+    "e": 24,
+    "f": 35,
+    "g": 31,
+    "h": 13,
+    "i": 4,
+    "j": 26,
+    "k": 5,
+    "l": 28,
+    "m": 15,
+    "n": 14,
+    "o": 17,
+    "p": 18,
+    "q": 37,
+    "r": 27,
+    "s": 12,
+    "t": 9,
+    "u": 25,
+    "v": 33,
+    "w": 8,
+    "x": 36,
+    "y": 16,
+    "z": 32,
+    "|": 7,
+    "ā": 11,
+    "ē": 10,
+    "ī": 19,
+    "ō": 23,
+    "‐": 20,
+    "—": 38
+  },
+  "crk-script_syllabics": {
+    "7": 81,
+    "9": 80,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "h": 67,
+    "t": 79,
+    "|": 4,
+    "ᐁ": 9,
+    "ᐃ": 7,
+    "ᐄ": 78,
+    "ᐅ": 23,
+    "ᐆ": 64,
+    "ᐊ": 8,
+    "ᐋ": 28,
+    "ᐟ": 30,
+    "ᐠ": 18,
+    "ᐢ": 15,
+    "ᐣ": 29,
+    "ᐤ": 27,
+    "ᐦ": 13,
+    "ᐧ": 5,
+    "ᐨ": 48,
+    "ᐩ": 60,
+    "ᐯ": 40,
+    "ᐱ": 34,
+    "ᐳ": 52,
+    "ᐴ": 76,
+    "ᐸ": 39,
+    "ᐹ": 59,
+    "ᑊ": 61,
+    "ᑌ": 36,
+    "ᑎ": 32,
+    "ᑐ": 37,
+    "ᑑ": 71,
+    "ᑕ": 12,
+    "ᑖ": 35,
+    "ᑦ": 53,
+    "ᑫ": 31,
+    "ᑭ": 6,
+    "ᑮ": 66,
+    "ᑯ": 21,
+    "ᑰ": 65,
+    "ᑲ": 11,
+    "ᑳ": 20,
+    "ᒉ": 56,
+    "ᒋ": 16,
+    "ᒌ": 77,
+    "ᒍ": 62,
+    "ᒐ": 63,
+    "ᒑ": 58,
+    "ᒣ": 47,
+    "ᒥ": 17,
+    "ᒦ": 75,
+    "ᒧ": 43,
+    "ᒨ": 69,
+    "ᒪ": 26,
+    "ᒫ": 24,
+    "ᒼ": 68,
+    "ᓀ": 49,
+    "ᓂ": 14,
+    "ᓃ": 72,
+    "ᓄ": 50,
+    "ᓅ": 73,
+    "ᓇ": 19,
+    "ᓈ": 41,
+    "ᓬ": 55,
+    "ᓭ": 46,
+    "ᓯ": 10,
+    "ᓰ": 74,
+    "ᓱ": 44,
+    "ᓲ": 70,
+    "ᓴ": 42,
+    "ᓵ": 54,
+    "ᔦ": 45,
+    "ᔨ": 22,
+    "ᔪ": 57,
+    "ᔭ": 25,
+    "ᔮ": 38,
+    "ᕒ": 51,
+    "ᕽ": 33
+  },
+  "crn": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 32,
+    "c": 17,
+    "d": 26,
+    "e": 9,
+    "f": 33,
+    "g": 34,
+    "h": 18,
+    "i": 10,
+    "j": 12,
+    "k": 38,
+    "l": 27,
+    "m": 14,
+    "n": 13,
+    "o": 24,
+    "p": 20,
+    "q": 31,
+    "r": 16,
+    "s": 19,
+    "t": 7,
+    "u": 8,
+    "v": 22,
+    "x": 21,
+    "y": 15,
+    "z": 23,
+    "|": 5,
+    "à": 39,
+    "á": 25,
+    "é": 29,
+    "í": 30,
+    "ñ": 37,
+    "ó": 36,
+    "ú": 28,
+    "ɨ": 11,
+    "́": 35
+  },
+  "crq": {
+    "'": 9,
+    "-": 48,
+    "0": 35,
+    "1": 37,
+    "2": 38,
+    "3": 45,
+    "4": 41,
+    "5": 42,
+    "6": 43,
+    "7": 40,
+    "8": 46,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 30,
+    "c": 27,
+    "d": 29,
+    "e": 7,
+    "f": 34,
+    "g": 31,
+    "h": 33,
+    "i": 6,
+    "j": 8,
+    "k": 14,
+    "l": 16,
+    "m": 19,
+    "n": 12,
+    "o": 13,
+    "p": 11,
+    "q": 39,
+    "r": 25,
+    "s": 18,
+    "t": 10,
+    "u": 21,
+    "v": 36,
+    "w": 15,
+    "x": 44,
+    "y": 17,
+    "z": 20,
+    "|": 4,
+    "á": 24,
+    "é": 28,
+    "í": 26,
+    "ñ": 49,
+    "ó": 32,
+    "ú": 23,
+    "–": 22
+  },
+  "crs": {
+    "'": 26,
+    "-": 27,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "3": 33,
+    "4": 32,
+    "5": 34,
+    "6": 35,
+    "7": 37,
+    "8": 38,
+    "9": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 21,
+    "d": 18,
+    "e": 7,
+    "f": 23,
+    "g": 25,
+    "h": 29,
+    "i": 9,
+    "k": 13,
+    "l": 14,
+    "m": 17,
+    "n": 5,
+    "o": 8,
+    "p": 15,
+    "r": 10,
+    "s": 12,
+    "t": 11,
+    "u": 19,
+    "v": 22,
+    "w": 24,
+    "y": 20,
+    "z": 16,
+    "|": 4
+  },
+  "crt": {
+    "'": 9,
+    "-": 39,
+    "0": 42,
+    "4": 40,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 27,
+    "c": 21,
+    "d": 25,
+    "e": 7,
+    "f": 32,
+    "g": 31,
+    "h": 22,
+    "i": 4,
+    "j": 8,
+    "k": 14,
+    "l": 17,
+    "m": 20,
+    "n": 10,
+    "o": 15,
+    "p": 16,
+    "q": 36,
+    "r": 23,
+    "s": 12,
+    "t": 11,
+    "u": 19,
+    "v": 35,
+    "w": 18,
+    "x": 37,
+    "y": 13,
+    "z": 34,
+    "|": 5,
+    "á": 30,
+    "é": 28,
+    "í": 29,
+    "ñ": 38,
+    "ó": 33,
+    "ú": 24,
+    "—": 26
+  },
+  "csk": {
+    "'": 35,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 32,
+    "d": 33,
+    "e": 12,
+    "f": 21,
+    "g": 29,
+    "h": 19,
+    "i": 8,
+    "j": 18,
+    "k": 10,
+    "l": 7,
+    "m": 13,
+    "n": 11,
+    "o": 9,
+    "s": 20,
+    "t": 15,
+    "u": 6,
+    "w": 17,
+    "y": 16,
+    "|": 4,
+    "á": 22,
+    "é": 26,
+    "í": 28,
+    "ñ": 25,
+    "ó": 30,
+    "ú": 23,
+    "ŋ": 24,
+    "ˈ": 31,
+    "̥": 27,
+    "—": 34
+  },
+  "cso": {
+    "1": 10,
+    "2": 6,
+    "3": 5,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 31,
+    "c": 17,
+    "d": 29,
+    "e": 25,
+    "f": 38,
+    "g": 34,
+    "h": 7,
+    "i": 12,
+    "j": 13,
+    "k": 40,
+    "l": 19,
+    "m": 21,
+    "n": 8,
+    "o": 23,
+    "p": 36,
+    "q": 27,
+    "r": 32,
+    "s": 18,
+    "t": 15,
+    "u": 11,
+    "v": 39,
+    "x": 41,
+    "y": 35,
+    "z": 33,
+    "|": 4,
+    "á": 9,
+    "é": 28,
+    "ë": 37,
+    "í": 20,
+    "ñ": 26,
+    "ó": 30,
+    "ú": 24,
+    "ɨ": 16,
+    "́": 22
+  },
+  "ctd": {
+    "'": 30,
+    "-": 24,
+    "0": 33,
+    "1": 35,
+    "2": 36,
+    "3": 40,
+    "4": 37,
+    "5": 43,
+    "6": 39,
+    "8": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 21,
+    "d": 19,
+    "e": 12,
+    "f": 29,
+    "g": 11,
+    "h": 8,
+    "i": 6,
+    "j": 27,
+    "k": 13,
+    "l": 15,
+    "m": 14,
+    "n": 7,
+    "o": 18,
+    "p": 16,
+    "q": 34,
+    "r": 26,
+    "s": 17,
+    "t": 9,
+    "u": 10,
+    "v": 25,
+    "w": 20,
+    "x": 32,
+    "y": 28,
+    "z": 22,
+    "|": 4,
+    "á": 41,
+    "–": 31,
+    "—": 38
+  },
+  "ctg": {
+    "'": 52,
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ঁ": 26,
+    "ং": 42,
+    "ঃ": 58,
+    "অ": 17,
+    "আ": 19,
+    "ই": 12,
+    "উ": 30,
+    "এ": 33,
+    "ঐ": 56,
+    "ও": 44,
+    "ক": 21,
+    "খ": 46,
+    "গ": 23,
+    "ঘ": 47,
+    "চ": 35,
+    "ছ": 32,
+    "জ": 29,
+    "ঝ": 45,
+    "ঞ": 59,
+    "ট": 40,
+    "ঠ": 48,
+    "ড": 38,
+    "ঢ": 43,
+    "ত": 13,
+    "থ": 39,
+    "দ": 24,
+    "ধ": 37,
+    "ন": 11,
+    "প": 57,
+    "ফ": 27,
+    "ব": 16,
+    "ভ": 36,
+    "ম": 20,
+    "য": 7,
+    "র": 6,
+    "ল": 15,
+    "শ": 31,
+    "ষ": 34,
+    "স": 28,
+    "হ": 18,
+    "়": 14,
+    "া": 5,
+    "ি": 9,
+    "ী": 60,
+    "ু": 22,
+    "ূ": 55,
+    "ৃ": 53,
+    "ে": 8,
+    "ৈ": 49,
+    "ো": 25,
+    "ৌ": 61,
+    "্": 10,
+    "‍": 50,
+    "–": 54,
+    "—": 51
+  },
+  "cto": {
+    "'": 9,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 7,
+    "c": 32,
+    "d": 14,
+    "e": 8,
+    "f": 39,
+    "g": 31,
+    "h": 26,
+    "i": 10,
+    "j": 24,
+    "k": 20,
+    "l": 35,
+    "m": 13,
+    "n": 21,
+    "o": 22,
+    "p": 29,
+    "q": 44,
+    "r": 6,
+    "s": 19,
+    "t": 18,
+    "u": 15,
+    "v": 42,
+    "w": 16,
+    "y": 30,
+    "z": 17,
+    "|": 5,
+    "á": 41,
+    "ã": 11,
+    "é": 38,
+    "í": 37,
+    "ñ": 36,
+    "ó": 40,
+    "õ": 34,
+    "ú": 33,
+    "ĩ": 27,
+    "ũ": 23,
+    "ʉ": 12,
+    "̃": 28,
+    "ẽ": 25
+  },
+  "ctu": {
+    "'": 15,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 14,
+    "c": 7,
+    "d": 28,
+    "e": 10,
+    "f": 35,
+    "g": 30,
+    "h": 20,
+    "i": 5,
+    "j": 12,
+    "k": 39,
+    "l": 8,
+    "m": 16,
+    "n": 13,
+    "o": 11,
+    "p": 23,
+    "q": 25,
+    "r": 27,
+    "s": 18,
+    "t": 9,
+    "u": 17,
+    "v": 37,
+    "w": 24,
+    "x": 26,
+    "y": 21,
+    "z": 36,
+    "|": 4,
+    "á": 32,
+    "é": 33,
+    "í": 31,
+    "ñ": 22,
+    "ó": 34,
+    "ú": 29,
+    "ń": 38,
+    "ʌ": 19
+  },
+  "cuc": {
+    "1": 19,
+    "2": 11,
+    "3": 5,
+    "4": 8,
+    "5": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 33,
+    "c": 17,
+    "d": 29,
+    "e": 12,
+    "f": 36,
+    "g": 24,
+    "h": 10,
+    "i": 6,
+    "j": 13,
+    "k": 42,
+    "l": 16,
+    "m": 21,
+    "n": 9,
+    "o": 14,
+    "p": 30,
+    "q": 23,
+    "r": 27,
+    "s": 20,
+    "t": 18,
+    "u": 15,
+    "v": 39,
+    "x": 41,
+    "y": 28,
+    "z": 40,
+    "|": 4,
+    "á": 35,
+    "é": 37,
+    "í": 34,
+    "ñ": 31,
+    "ó": 38,
+    "ú": 26,
+    "ŋ": 32,
+    "ˉ": 22
+  },
+  "cui": {
+    "'": 33,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 13,
+    "c": 12,
+    "d": 24,
+    "e": 6,
+    "f": 29,
+    "g": 32,
+    "h": 22,
+    "i": 8,
+    "j": 21,
+    "k": 39,
+    "l": 27,
+    "m": 16,
+    "n": 7,
+    "o": 9,
+    "p": 10,
+    "q": 26,
+    "r": 20,
+    "s": 17,
+    "t": 11,
+    "u": 15,
+    "v": 34,
+    "w": 18,
+    "x": 14,
+    "y": 19,
+    "z": 36,
+    "|": 5,
+    "á": 37,
+    "é": 28,
+    "í": 31,
+    "ñ": 40,
+    "ó": 35,
+    "ú": 25,
+    "ʉ": 23,
+    "—": 30
+  },
+  "cuk": {
+    "'": 38,
+    "-": 13,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "c": 24,
+    "d": 7,
+    "e": 6,
+    "f": 30,
+    "g": 9,
+    "h": 27,
+    "i": 8,
+    "j": 23,
+    "k": 19,
+    "l": 18,
+    "m": 12,
+    "n": 10,
+    "o": 17,
+    "p": 25,
+    "q": 36,
+    "r": 15,
+    "s": 16,
+    "t": 22,
+    "u": 14,
+    "v": 33,
+    "w": 20,
+    "x": 37,
+    "y": 21,
+    "z": 35,
+    "|": 5,
+    "á": 34,
+    "é": 28,
+    "í": 31,
+    "ó": 32,
+    "ú": 26,
+    "—": 29
+  },
+  "cul": {
+    "'": 43,
+    "-": 23,
+    "0": 25,
+    "1": 26,
+    "2": 29,
+    "3": 30,
+    "4": 31,
+    "5": 33,
+    "6": 36,
+    "7": 28,
+    "8": 39,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 15,
+    "d": 9,
+    "e": 10,
+    "f": 27,
+    "g": 34,
+    "h": 16,
+    "i": 5,
+    "j": 11,
+    "k": 40,
+    "l": 24,
+    "m": 19,
+    "n": 7,
+    "o": 8,
+    "p": 18,
+    "q": 20,
+    "r": 14,
+    "s": 12,
+    "t": 13,
+    "u": 17,
+    "v": 32,
+    "z": 37,
+    "|": 6,
+    "ñ": 42,
+    "ó": 41,
+    "ʼ": 35,
+    "—": 22
+  },
+  "cwa": {
+    "'": 28,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 13,
+    "c": 29,
+    "d": 22,
+    "e": 7,
+    "f": 27,
+    "g": 17,
+    "h": 11,
+    "i": 9,
+    "j": 23,
+    "k": 14,
+    "m": 16,
+    "n": 10,
+    "o": 12,
+    "p": 26,
+    "r": 8,
+    "s": 20,
+    "t": 19,
+    "u": 6,
+    "v": 30,
+    "w": 18,
+    "y": 15,
+    "|": 5,
+    "ú": 21,
+    "ʼ": 24
+  },
+  "cwe": {
+    "'": 26,
+    "-": 29,
+    "0": 30,
+    "2": 32,
+    "3": 33,
+    "7": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 24,
+    "d": 19,
+    "e": 10,
+    "f": 25,
+    "g": 16,
+    "h": 15,
+    "i": 6,
+    "j": 27,
+    "k": 14,
+    "l": 9,
+    "m": 13,
+    "n": 8,
+    "o": 12,
+    "p": 28,
+    "s": 18,
+    "t": 22,
+    "u": 7,
+    "v": 23,
+    "w": 11,
+    "y": 17,
+    "z": 20,
+    "|": 5
+  },
+  "cwt": {
+    "'": 19,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 26,
+    "e": 6,
+    "f": 22,
+    "g": 32,
+    "h": 16,
+    "i": 7,
+    "j": 21,
+    "k": 10,
+    "l": 25,
+    "m": 15,
+    "n": 8,
+    "o": 11,
+    "p": 34,
+    "r": 14,
+    "s": 18,
+    "t": 12,
+    "u": 9,
+    "w": 24,
+    "y": 17,
+    "|": 4,
+    "á": 23,
+    "é": 30,
+    "í": 29,
+    "ñ": 27,
+    "ó": 35,
+    "ú": 28,
+    "ŋ": 20,
+    "̱": 33
+  },
+  "cya": {
+    "'": 7,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 10,
+    "d": 19,
+    "e": 15,
+    "f": 32,
+    "g": 23,
+    "h": 16,
+    "i": 9,
+    "j": 21,
+    "k": 38,
+    "l": 17,
+    "m": 20,
+    "n": 6,
+    "o": 11,
+    "p": 27,
+    "q": 24,
+    "r": 25,
+    "s": 14,
+    "t": 12,
+    "u": 8,
+    "v": 33,
+    "x": 22,
+    "y": 13,
+    "z": 35,
+    "|": 4,
+    "á": 36,
+    "é": 31,
+    "í": 30,
+    "ñ": 18,
+    "ó": 34,
+    "ú": 28,
+    "ü": 37,
+    "—": 29
+  },
+  "cym": {
+    "\"": 67,
+    "$": 55,
+    "%": 58,
+    "'": 24,
+    "+": 66,
+    ",": 44,
+    "-": 32,
+    ".": 41,
+    "/": 54,
+    "0": 26,
+    "1": 28,
+    "2": 31,
+    "3": 38,
+    "4": 37,
+    "5": 35,
+    "6": 39,
+    "7": 43,
+    "8": 40,
+    "9": 33,
+    ":": 52,
+    ";": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 77,
+    "]": 78,
+    "a": 5,
+    "b": 23,
+    "c": 21,
+    "d": 6,
+    "e": 8,
+    "f": 17,
+    "g": 16,
+    "h": 15,
+    "i": 10,
+    "j": 34,
+    "k": 29,
+    "l": 13,
+    "m": 22,
+    "n": 7,
+    "o": 12,
+    "p": 25,
+    "q": 51,
+    "r": 11,
+    "s": 19,
+    "t": 18,
+    "u": 20,
+    "v": 36,
+    "w": 14,
+    "x": 48,
+    "y": 9,
+    "z": 45,
+    "|": 4,
+    "£": 69,
+    "¥": 61,
+    "°": 76,
+    "à": 80,
+    "á": 50,
+    "â": 27,
+    "ã": 71,
+    "ä": 73,
+    "ç": 68,
+    "é": 56,
+    "ê": 47,
+    "ë": 63,
+    "ì": 72,
+    "í": 60,
+    "î": 49,
+    "ï": 46,
+    "ò": 70,
+    "ó": 74,
+    "ô": 30,
+    "õ": 62,
+    "ö": 64,
+    "ú": 65,
+    "û": 79,
+    "ü": 57,
+    "ŵ": 42,
+    "ŷ": 53,
+    "̇": 75,
+    "–": 81,
+    "—": 82
+  },
+  "daa": {
+    "'": 33,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 24,
+    "d": 21,
+    "e": 11,
+    "g": 14,
+    "h": 29,
+    "i": 6,
+    "j": 28,
+    "k": 7,
+    "l": 20,
+    "m": 17,
+    "n": 8,
+    "o": 9,
+    "p": 26,
+    "r": 12,
+    "s": 18,
+    "t": 13,
+    "u": 10,
+    "w": 23,
+    "y": 16,
+    "z": 30,
+    "|": 4,
+    "ŋ": 15,
+    "ƴ": 32,
+    "ɓ": 31,
+    "ɗ": 22,
+    "̰": 27
+  },
+  "dah": {
+    "'": 36,
+    "-": 37,
+    "0": 26,
+    "1": 27,
+    "2": 29,
+    "3": 32,
+    "4": 30,
+    "5": 33,
+    "6": 31,
+    "7": 28,
+    "8": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 15,
+    "d": 18,
+    "e": 5,
+    "f": 25,
+    "g": 13,
+    "h": 20,
+    "i": 8,
+    "k": 17,
+    "l": 21,
+    "m": 16,
+    "n": 12,
+    "o": 14,
+    "p": 24,
+    "r": 9,
+    "s": 22,
+    "t": 19,
+    "u": 10,
+    "w": 23,
+    "y": 11,
+    "|": 4,
+    "ŋ": 7,
+    "–": 38
+  },
+  "dan": {
+    "\"": 57,
+    "$": 51,
+    "%": 53,
+    "&": 64,
+    "'": 45,
+    "+": 73,
+    ",": 49,
+    "-": 32,
+    ".": 33,
+    "/": 46,
+    "0": 30,
+    "1": 31,
+    "2": 35,
+    "3": 39,
+    "4": 38,
+    "5": 37,
+    "6": 41,
+    "7": 43,
+    "8": 42,
+    "9": 36,
+    ":": 50,
+    ";": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 68,
+    "]": 69,
+    "a": 9,
+    "b": 21,
+    "c": 29,
+    "d": 10,
+    "e": 5,
+    "f": 18,
+    "g": 15,
+    "h": 23,
+    "i": 11,
+    "j": 28,
+    "k": 17,
+    "l": 13,
+    "m": 16,
+    "n": 7,
+    "o": 14,
+    "p": 22,
+    "q": 48,
+    "r": 6,
+    "s": 12,
+    "t": 8,
+    "u": 20,
+    "v": 19,
+    "w": 34,
+    "x": 44,
+    "y": 26,
+    "z": 40,
+    "|": 4,
+    "¥": 63,
+    "°": 61,
+    "²": 72,
+    "á": 70,
+    "ã": 67,
+    "å": 24,
+    "æ": 25,
+    "ç": 66,
+    "é": 47,
+    "í": 54,
+    "ó": 74,
+    "õ": 58,
+    "ö": 62,
+    "ø": 27,
+    "ü": 55,
+    "ł": 71,
+    "​": 52,
+    "–": 59,
+    "—": 60,
+    "”": 65
+  },
+  "dar": {
+    "'": 38,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 6,
+    "а": 5,
+    "б": 9,
+    "в": 21,
+    "г": 18,
+    "д": 15,
+    "е": 13,
+    "ж": 31,
+    "з": 27,
+    "и": 4,
+    "й": 29,
+    "к": 17,
+    "л": 7,
+    "м": 24,
+    "н": 14,
+    "о": 36,
+    "п": 30,
+    "р": 8,
+    "с": 16,
+    "т": 19,
+    "у": 10,
+    "ф": 35,
+    "х": 12,
+    "ц": 28,
+    "ч": 23,
+    "ш": 26,
+    "ъ": 22,
+    "ь": 25,
+    "э": 33,
+    "ю": 37,
+    "я": 20,
+    "ӏ": 11,
+    "–": 32
+  },
+  "dbj": {
+    "'": 20,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 29,
+    "d": 21,
+    "e": 10,
+    "f": 27,
+    "g": 12,
+    "h": 26,
+    "i": 9,
+    "j": 23,
+    "k": 14,
+    "l": 17,
+    "m": 11,
+    "n": 6,
+    "o": 8,
+    "p": 19,
+    "r": 16,
+    "s": 13,
+    "t": 15,
+    "u": 7,
+    "w": 24,
+    "y": 22,
+    "z": 28,
+    "|": 4
+  },
+  "dbq": {
+    "'": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 28,
+    "d": 16,
+    "e": 13,
+    "f": 31,
+    "g": 18,
+    "h": 20,
+    "i": 6,
+    "j": 27,
+    "k": 8,
+    "l": 12,
+    "m": 11,
+    "n": 7,
+    "o": 22,
+    "p": 29,
+    "r": 14,
+    "s": 24,
+    "t": 9,
+    "u": 17,
+    "v": 25,
+    "w": 21,
+    "y": 19,
+    "z": 15,
+    "|": 4,
+    "á": 32,
+    "é": 35,
+    "ú": 33,
+    "ɓ": 30,
+    "ɗ": 23,
+    "ə": 10
+  },
+  "ddn": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 32,
+    "b": 15,
+    "c": 24,
+    "d": 14,
+    "e": 13,
+    "f": 27,
+    "g": 17,
+    "h": 21,
+    "i": 6,
+    "j": 31,
+    "k": 9,
+    "l": 23,
+    "m": 8,
+    "n": 7,
+    "o": 10,
+    "p": 28,
+    "r": 20,
+    "s": 18,
+    "t": 22,
+    "u": 11,
+    "w": 25,
+    "y": 19,
+    "z": 30,
+    "|": 4,
+    "à": 33,
+    "á": 38,
+    "ã": 35,
+    "ò": 37,
+    "ŋ": 34,
+    "ǹ": 29,
+    "ɑ": 5,
+    "ɔ": 16,
+    "ɛ": 12,
+    "̀": 26,
+    "̃": 36
+  },
+  "ded": {
+    "'": 32,
+    "0": 27,
+    "1": 26,
+    "2": 28,
+    "3": 35,
+    "4": 30,
+    "5": 33,
+    "6": 34,
+    "7": 29,
+    "8": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 9,
+    "d": 14,
+    "e": 5,
+    "f": 19,
+    "g": 8,
+    "h": 21,
+    "i": 10,
+    "k": 15,
+    "l": 31,
+    "m": 11,
+    "n": 7,
+    "o": 12,
+    "p": 23,
+    "r": 17,
+    "s": 18,
+    "t": 22,
+    "u": 13,
+    "w": 24,
+    "y": 16,
+    "z": 25,
+    "|": 4,
+    "—": 38
+  },
+  "des": {
+    "-": 44,
+    "1": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 19,
+    "d": 22,
+    "e": 7,
+    "f": 38,
+    "g": 12,
+    "h": 17,
+    "i": 8,
+    "j": 21,
+    "k": 45,
+    "l": 33,
+    "m": 11,
+    "n": 26,
+    "o": 10,
+    "p": 15,
+    "q": 31,
+    "r": 5,
+    "s": 18,
+    "t": 23,
+    "u": 14,
+    "v": 39,
+    "w": 25,
+    "x": 46,
+    "y": 24,
+    "z": 41,
+    "|": 4,
+    "á": 27,
+    "ã": 13,
+    "é": 37,
+    "í": 36,
+    "ñ": 28,
+    "ó": 35,
+    "õ": 30,
+    "ú": 43,
+    "ü": 40,
+    "ĩ": 16,
+    "ũ": 34,
+    "ʉ": 9,
+    "̃": 29,
+    "̱": 48,
+    "ẽ": 32,
+    "‐": 47
+  },
+  "deu": {
+    "!": 70,
+    "\"": 39,
+    "$": 75,
+    "%": 53,
+    "&": 57,
+    "'": 51,
+    "+": 58,
+    ",": 50,
+    "-": 32,
+    ".": 36,
+    "/": 49,
+    "0": 31,
+    "1": 35,
+    "2": 37,
+    "3": 42,
+    "4": 43,
+    "5": 41,
+    "6": 44,
+    "7": 47,
+    "8": 45,
+    "9": 40,
+    ":": 48,
+    ";": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 62,
+    "]": 63,
+    "`": 126,
+    "a": 11,
+    "b": 20,
+    "c": 18,
+    "d": 12,
+    "e": 5,
+    "f": 21,
+    "g": 16,
+    "h": 13,
+    "i": 7,
+    "j": 30,
+    "k": 23,
+    "l": 15,
+    "m": 19,
+    "n": 6,
+    "o": 17,
+    "p": 26,
+    "q": 46,
+    "r": 8,
+    "s": 9,
+    "t": 10,
+    "u": 14,
+    "v": 25,
+    "w": 22,
+    "x": 38,
+    "y": 33,
+    "z": 24,
+    "|": 4,
+    "¥": 65,
+    "§": 182,
+    "°": 54,
+    "²": 64,
+    "ß": 34,
+    "à": 130,
+    "á": 52,
+    "â": 83,
+    "ã": 69,
+    "ä": 28,
+    "å": 108,
+    "æ": 66,
+    "ç": 60,
+    "è": 177,
+    "é": 56,
+    "ê": 107,
+    "ë": 95,
+    "ì": 132,
+    "í": 55,
+    "î": 111,
+    "ï": 117,
+    "ð": 110,
+    "ñ": 90,
+    "ò": 118,
+    "ó": 76,
+    "ô": 88,
+    "õ": 71,
+    "ö": 29,
+    "ø": 84,
+    "ù": 188,
+    "ú": 67,
+    "û": 144,
+    "ü": 27,
+    "ý": 89,
+    "þ": 183,
+    "ā": 68,
+    "ă": 109,
+    "ą": 112,
+    "ć": 82,
+    "č": 80,
+    "ď": 134,
+    "đ": 103,
+    "ē": 185,
+    "ė": 116,
+    "ę": 124,
+    "ě": 96,
+    "ğ": 114,
+    "ġ": 133,
+    "ħ": 170,
+    "ī": 100,
+    "ı": 86,
+    "ł": 72,
+    "ń": 105,
+    "ņ": 161,
+    "ň": 113,
+    "ō": 78,
+    "ŏ": 165,
+    "ő": 123,
+    "œ": 98,
+    "ř": 81,
+    "ś": 97,
+    "ş": 87,
+    "š": 79,
+    "ť": 135,
+    "ū": 91,
+    "ů": 122,
+    "ź": 121,
+    "ż": 119,
+    "ž": 85,
+    "ơ": 127,
+    "ǐ": 152,
+    "ș": 92,
+    "ț": 99,
+    "ə": 101,
+    "ʻ": 102,
+    "ʾ": 164,
+    "ʿ": 94,
+    "́": 93,
+    "̆": 163,
+    "̇": 115,
+    "̥": 178,
+    "μ": 145,
+    "а": 128,
+    "в": 142,
+    "е": 138,
+    "и": 129,
+    "к": 160,
+    "м": 141,
+    "о": 125,
+    "р": 139,
+    "с": 137,
+    "ф": 140,
+    "ч": 159,
+    "ш": 143,
+    "ѹ": 150,
+    "་": 154,
+    "ན": 153,
+    "ḫ": 120,
+    "ṟ": 184,
+    "ṣ": 146,
+    "ṭ": 190,
+    "ạ": 148,
+    "ả": 147,
+    "ắ": 172,
+    "ằ": 193,
+    "ế": 136,
+    "ễ": 192,
+    "ộ": 191,
+    "ụ": 171,
+    "‌": 73,
+    "‍": 61,
+    "‐": 131,
+    "–": 77,
+    "—": 104,
+    "”": 74,
+    "′": 106,
+    "≡": 189,
+    "⟨": 186,
+    "⟩": 187,
+    "カ": 155,
+    "东": 181,
+    "临": 168,
+    "关": 175,
+    "合": 162,
+    "城": 174,
+    "孙": 166,
+    "尣": 169,
+    "支": 158,
+    "比": 157,
+    "毛": 179,
+    "泽": 180,
+    "無": 151,
+    "臣": 156,
+    "辶": 149,
+    "道": 167,
+    "镇": 176,
+    "黃": 173
+  },
+  "dga": {
+    "'": 37,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "d": 19,
+    "e": 6,
+    "f": 27,
+    "g": 25,
+    "h": 34,
+    "i": 15,
+    "k": 14,
+    "l": 11,
+    "m": 18,
+    "n": 7,
+    "o": 8,
+    "p": 23,
+    "r": 16,
+    "s": 22,
+    "t": 20,
+    "u": 21,
+    "v": 28,
+    "w": 26,
+    "y": 13,
+    "z": 24,
+    "|": 4,
+    "ã": 29,
+    "õ": 33,
+    "ĩ": 36,
+    "ŋ": 10,
+    "ũ": 32,
+    "ɔ": 17,
+    "ɛ": 9,
+    "̃": 31,
+    "ẽ": 35
+  },
+  "dgi": {
+    "'": 32,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 34,
+    "d": 29,
+    "e": 14,
+    "f": 33,
+    "g": 38,
+    "h": 48,
+    "i": 22,
+    "j": 49,
+    "k": 19,
+    "l": 13,
+    "m": 16,
+    "n": 7,
+    "o": 28,
+    "p": 26,
+    "r": 11,
+    "s": 23,
+    "t": 12,
+    "u": 27,
+    "v": 36,
+    "w": 15,
+    "y": 17,
+    "z": 24,
+    "|": 4,
+    "à": 45,
+    "á": 42,
+    "ã": 25,
+    "è": 55,
+    "é": 50,
+    "ì": 52,
+    "í": 46,
+    "ò": 54,
+    "ó": 53,
+    "õ": 39,
+    "ù": 44,
+    "ú": 43,
+    "ĩ": 51,
+    "ŋ": 30,
+    "ũ": 41,
+    "ƴ": 31,
+    "ɓ": 47,
+    "ɔ": 18,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ʋ": 10,
+    "̀": 37,
+    "́": 35,
+    "̃": 20,
+    "ẽ": 40
+  },
+  "dgk": {
+    "'": 17,
+    "-": 37,
+    "2": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "d": 12,
+    "e": 13,
+    "g": 15,
+    "h": 26,
+    "i": 10,
+    "j": 23,
+    "k": 8,
+    "l": 16,
+    "m": 18,
+    "n": 11,
+    "o": 7,
+    "p": 32,
+    "r": 19,
+    "s": 27,
+    "t": 9,
+    "u": 14,
+    "w": 35,
+    "y": 24,
+    "|": 4,
+    "à": 45,
+    "á": 36,
+    "é": 30,
+    "ì": 44,
+    "í": 48,
+    "ò": 42,
+    "ó": 43,
+    "ù": 40,
+    "ú": 39,
+    "ŋ": 21,
+    "ɓ": 29,
+    "ɔ": 20,
+    "ɗ": 25,
+    "ə": 46,
+    "ɛ": 22,
+    "̀": 41,
+    "́": 33,
+    "̰": 6,
+    "ә": 38,
+    "ḭ": 31,
+    "ṵ": 34
+  },
+  "dgo": {
+    "'": 63,
+    "-": 53,
+    "4": 66,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 46,
+    "ं": 14,
+    "ः": 62,
+    "अ": 36,
+    "आ": 28,
+    "इ": 49,
+    "ई": 30,
+    "उ": 33,
+    "ऊ": 44,
+    "ए": 37,
+    "ऐ": 35,
+    "ओ": 27,
+    "औ": 58,
+    "क": 9,
+    "ख": 32,
+    "ग": 21,
+    "घ": 59,
+    "च": 26,
+    "छ": 45,
+    "ज": 22,
+    "झ": 54,
+    "ञ": 60,
+    "ट": 43,
+    "ठ": 52,
+    "ड": 31,
+    "ढ": 55,
+    "ण": 57,
+    "त": 12,
+    "थ": 41,
+    "द": 10,
+    "ध": 51,
+    "न": 7,
+    "प": 17,
+    "फ": 47,
+    "ब": 25,
+    "भ": 50,
+    "म": 18,
+    "य": 20,
+    "र": 8,
+    "ल": 16,
+    "व": 39,
+    "श": 38,
+    "ष": 56,
+    "स": 13,
+    "ह": 40,
+    "़": 34,
+    "ा": 6,
+    "ि": 19,
+    "ी": 11,
+    "ु": 24,
+    "ू": 42,
+    "ृ": 61,
+    "े": 5,
+    "ै": 23,
+    "ॉ": 64,
+    "ो": 29,
+    "ौ": 48,
+    "्": 15,
+    "‍": 65
+  },
+  "dgr": {
+    "'": 11,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 40,
+    "c": 33,
+    "d": 12,
+    "e": 5,
+    "f": 44,
+    "g": 14,
+    "h": 8,
+    "i": 42,
+    "j": 37,
+    "k": 22,
+    "l": 25,
+    "m": 36,
+    "n": 16,
+    "o": 21,
+    "p": 38,
+    "q": 45,
+    "r": 35,
+    "s": 13,
+    "t": 10,
+    "u": 39,
+    "v": 43,
+    "w": 23,
+    "x": 27,
+    "y": 24,
+    "z": 26,
+    "|": 4,
+    "à": 18,
+    "è": 20,
+    "ì": 17,
+    "ò": 30,
+    "ą": 41,
+    "ę": 34,
+    "į": 31,
+    "ı": 6,
+    "ł": 28,
+    "ǫ": 9,
+    "ɂ": 32,
+    "ʔ": 46,
+    "̀": 19,
+    "̨": 15
+  },
+  "dhi": {
+    "'": 52,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ँ": 48,
+    "ं": 58,
+    "अ": 30,
+    "आ": 32,
+    "इ": 14,
+    "उ": 26,
+    "ए": 38,
+    "ऐ": 54,
+    "ओ": 40,
+    "औ": 53,
+    "क": 7,
+    "ख": 28,
+    "ग": 25,
+    "घ": 36,
+    "ङ": 12,
+    "च": 33,
+    "छ": 45,
+    "ज": 29,
+    "झ": 42,
+    "ट": 41,
+    "ठ": 47,
+    "ड": 44,
+    "ढ": 49,
+    "ण": 55,
+    "त": 18,
+    "थ": 34,
+    "द": 22,
+    "ध": 35,
+    "न": 16,
+    "प": 19,
+    "फ": 39,
+    "ब": 23,
+    "भ": 31,
+    "म": 21,
+    "य": 24,
+    "र": 17,
+    "ल": 10,
+    "व": 27,
+    "श": 56,
+    "स": 15,
+    "ह": 13,
+    "़": 57,
+    "ा": 4,
+    "ि": 6,
+    "ु": 20,
+    "ृ": 51,
+    "े": 8,
+    "ै": 46,
+    "ो": 11,
+    "ौ": 50,
+    "्": 9,
+    "‍": 37
+  },
+  "did": {
+    "'": 34,
+    "-": 35,
+    "1": 36,
+    "6": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 13,
+    "d": 25,
+    "e": 15,
+    "g": 11,
+    "h": 9,
+    "i": 10,
+    "j": 28,
+    "k": 20,
+    "l": 19,
+    "m": 21,
+    "n": 7,
+    "o": 16,
+    "p": 27,
+    "r": 23,
+    "s": 31,
+    "t": 8,
+    "u": 17,
+    "v": 29,
+    "w": 30,
+    "y": 22,
+    "|": 4,
+    "á": 32,
+    "õ": 14,
+    "ú": 33,
+    "ĩ": 6,
+    "ũ": 26,
+    "ẽ": 12,
+    "ꞌ": 18
+  },
+  "dig": {
+    "'": 29,
+    "-": 30,
+    "0": 33,
+    "5": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 25,
+    "d": 19,
+    "e": 12,
+    "f": 28,
+    "g": 20,
+    "h": 10,
+    "i": 6,
+    "j": 27,
+    "k": 11,
+    "l": 14,
+    "m": 9,
+    "n": 7,
+    "o": 13,
+    "p": 22,
+    "r": 17,
+    "s": 18,
+    "t": 21,
+    "u": 8,
+    "v": 26,
+    "w": 15,
+    "y": 16,
+    "z": 23,
+    "|": 5,
+    "á": 31,
+    "ó": 32
+  },
+  "dik": {
+    "'": 34,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 23,
+    "c": 9,
+    "d": 28,
+    "e": 10,
+    "g": 31,
+    "h": 17,
+    "i": 8,
+    "j": 32,
+    "k": 5,
+    "l": 14,
+    "m": 24,
+    "n": 7,
+    "o": 22,
+    "p": 27,
+    "r": 20,
+    "t": 12,
+    "u": 11,
+    "w": 26,
+    "y": 18,
+    "|": 4,
+    "ä": 25,
+    "ë": 13,
+    "ï": 15,
+    "ö": 30,
+    "ŋ": 29,
+    "ɔ": 16,
+    "ɛ": 21,
+    "ɣ": 33,
+    "̈": 19
+  },
+  "dip": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 22,
+    "c": 12,
+    "d": 19,
+    "e": 5,
+    "g": 26,
+    "h": 18,
+    "i": 6,
+    "j": 27,
+    "k": 7,
+    "l": 16,
+    "m": 21,
+    "n": 9,
+    "o": 14,
+    "p": 24,
+    "r": 17,
+    "t": 11,
+    "u": 10,
+    "w": 23,
+    "y": 15,
+    "|": 4,
+    "ı": 31,
+    "ŋ": 25,
+    "ɔ": 13,
+    "ɛ": 20,
+    "ɣ": 28,
+    "ԑ": 29,
+    "—": 30
+  },
+  "div": {
+    "'": 64,
+    "-": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ا": 47,
+    "س": 59,
+    "ص": 54,
+    "ع": 56,
+    "ل": 44,
+    "م": 60,
+    "ه": 46,
+    "و": 58,
+    "ى": 55,
+    "ي": 57,
+    "ހ": 17,
+    "ށ": 25,
+    "ނ": 8,
+    "ރ": 14,
+    "ބ": 26,
+    "ޅ": 29,
+    "ކ": 15,
+    "އ": 7,
+    "ވ": 16,
+    "މ": 13,
+    "ފ": 23,
+    "ދ": 20,
+    "ތ": 19,
+    "ލ": 24,
+    "ގ": 18,
+    "ޏ": 49,
+    "ސ": 22,
+    "ޑ": 31,
+    "ޒ": 36,
+    "ޓ": 33,
+    "ޔ": 30,
+    "ޕ": 38,
+    "ޖ": 32,
+    "ޗ": 39,
+    "ޘ": 51,
+    "ޙ": 40,
+    "ޚ": 43,
+    "ޛ": 61,
+    "ޝ": 37,
+    "ޞ": 45,
+    "ޟ": 50,
+    "ޠ": 48,
+    "ޡ": 63,
+    "ޢ": 41,
+    "ޣ": 52,
+    "ޤ": 42,
+    "ޥ": 62,
+    "ަ": 5,
+    "ާ": 12,
+    "ި": 11,
+    "ީ": 21,
+    "ު": 10,
+    "ޫ": 35,
+    "ެ": 9,
+    "ޭ": 27,
+    "ޮ": 28,
+    "ޯ": 34,
+    "ް": 6
+  },
+  "djk": {
+    "-": 31,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "d": 12,
+    "e": 7,
+    "f": 16,
+    "g": 18,
+    "h": 25,
+    "i": 6,
+    "j": 24,
+    "k": 13,
+    "l": 20,
+    "m": 15,
+    "n": 8,
+    "o": 10,
+    "p": 21,
+    "r": 32,
+    "s": 11,
+    "t": 14,
+    "u": 9,
+    "v": 26,
+    "w": 19,
+    "y": 22,
+    "z": 27,
+    "|": 4,
+    "á": 23
+  },
+  "dnj-dialect_blowowest": {
+    "'": 7,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 15,
+    "d": 13,
+    "e": 31,
+    "f": 32,
+    "g": 20,
+    "h": 10,
+    "i": 23,
+    "k": 11,
+    "l": 25,
+    "m": 27,
+    "n": 9,
+    "o": 17,
+    "p": 26,
+    "r": 34,
+    "s": 28,
+    "t": 30,
+    "u": 16,
+    "v": 33,
+    "w": 19,
+    "y": 18,
+    "z": 29,
+    "|": 4,
+    "ë": 12,
+    "ö": 8,
+    "ü": 24,
+    "ɔ": 22,
+    "ɛ": 14,
+    "‐": 5,
+    "꞊": 21
+  },
+  "dnj-dialect_gweetaawueast": {
+    "'": 7,
+    "1": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 14,
+    "d": 11,
+    "e": 32,
+    "f": 35,
+    "g": 16,
+    "h": 8,
+    "i": 21,
+    "k": 12,
+    "l": 33,
+    "m": 25,
+    "n": 9,
+    "o": 20,
+    "p": 27,
+    "r": 36,
+    "s": 26,
+    "t": 28,
+    "u": 23,
+    "v": 37,
+    "w": 19,
+    "y": 15,
+    "z": 30,
+    "|": 4,
+    "ë": 17,
+    "ö": 10,
+    "ü": 24,
+    "ɔ": 18,
+    "ɛ": 13,
+    "ɩ": 34,
+    "ʋ": 29,
+    "̈": 31,
+    "‐": 5,
+    "꞊": 22
+  },
+  "dnt": {
+    "'": 24,
+    "-": 25,
+    "0": 27,
+    "1": 28,
+    "2": 31,
+    "3": 33,
+    "4": 32,
+    "5": 35,
+    "6": 36,
+    "7": 29,
+    "8": 39,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 38,
+    "d": 21,
+    "e": 6,
+    "f": 26,
+    "g": 12,
+    "h": 11,
+    "i": 7,
+    "j": 34,
+    "k": 8,
+    "l": 14,
+    "m": 16,
+    "n": 10,
+    "o": 9,
+    "p": 23,
+    "r": 18,
+    "s": 17,
+    "t": 13,
+    "u": 15,
+    "v": 40,
+    "w": 20,
+    "y": 22,
+    "z": 30,
+    "|": 4
+  },
+  "dnw": {
+    "'": 22,
+    "-": 23,
+    "0": 28,
+    "1": 25,
+    "2": 29,
+    "3": 34,
+    "4": 30,
+    "5": 31,
+    "6": 33,
+    "7": 27,
+    "8": 36,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 37,
+    "d": 21,
+    "e": 6,
+    "g": 9,
+    "h": 26,
+    "i": 11,
+    "j": 32,
+    "k": 10,
+    "l": 17,
+    "m": 12,
+    "n": 7,
+    "o": 8,
+    "p": 19,
+    "r": 13,
+    "s": 24,
+    "t": 14,
+    "u": 15,
+    "w": 16,
+    "y": 20,
+    "|": 4
+  },
+  "dop": {
+    "-": 27,
+    "3": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "c": 30,
+    "e": 22,
+    "f": 31,
+    "h": 29,
+    "i": 19,
+    "k": 16,
+    "l": 12,
+    "m": 17,
+    "n": 8,
+    "o": 28,
+    "p": 11,
+    "s": 15,
+    "t": 6,
+    "u": 23,
+    "w": 24,
+    "y": 21,
+    "|": 4,
+    "á": 26,
+    "é": 35,
+    "í": 36,
+    "ú": 34,
+    "ń": 33,
+    "ŋ": 25,
+    "ɔ": 9,
+    "ə": 14,
+    "ɛ": 10,
+    "ɣ": 18,
+    "ɩ": 7,
+    "ʋ": 13,
+    "́": 20,
+    "ḿ": 32
+  },
+  "dos": {
+    "0": 45,
+    "1": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 42,
+    "d": 27,
+    "e": 14,
+    "f": 36,
+    "g": 20,
+    "h": 13,
+    "i": 11,
+    "k": 10,
+    "l": 24,
+    "m": 16,
+    "n": 7,
+    "o": 28,
+    "p": 26,
+    "r": 9,
+    "s": 18,
+    "t": 25,
+    "u": 22,
+    "v": 38,
+    "w": 19,
+    "y": 15,
+    "z": 29,
+    "|": 4,
+    "à": 41,
+    "á": 43,
+    "ã": 31,
+    "í": 46,
+    "ĩ": 33,
+    "ŋ": 34,
+    "ũ": 37,
+    "ɓ": 30,
+    "ɔ": 23,
+    "ɗ": 32,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ʋ": 12,
+    "ʔ": 35,
+    "̀": 40,
+    "́": 39,
+    "̃": 21
+  },
+  "dsh": {
+    "'": 40,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 28,
+    "d": 16,
+    "e": 6,
+    "f": 29,
+    "g": 12,
+    "h": 10,
+    "i": 7,
+    "j": 33,
+    "k": 11,
+    "l": 9,
+    "m": 15,
+    "n": 13,
+    "o": 18,
+    "r": 23,
+    "s": 24,
+    "t": 22,
+    "u": 20,
+    "v": 34,
+    "w": 27,
+    "y": 14,
+    "|": 4,
+    "á": 8,
+    "â": 38,
+    "é": 21,
+    "ê": 39,
+    "í": 17,
+    "î": 37,
+    "ó": 32,
+    "ô": 36,
+    "ú": 25,
+    "û": 42,
+    "đ": 41,
+    "ʼ": 35,
+    "ˈ": 19,
+    "̶": 30
+  },
+  "dso": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଁ": 32,
+    "ଅ": 25,
+    "ଆ": 22,
+    "ଇ": 12,
+    "ଉ": 29,
+    "ଏ": 28,
+    "କ": 10,
+    "ଗ": 23,
+    "ଙ": 30,
+    "ଚ": 26,
+    "ଜ": 19,
+    "ଞ": 35,
+    "ଟ": 24,
+    "ଡ": 27,
+    "ଣ": 33,
+    "ତ": 17,
+    "ଦ": 21,
+    "ନ": 13,
+    "ପ": 20,
+    "ବ": 16,
+    "ମ": 15,
+    "ର": 9,
+    "ଲ": 14,
+    "ସ": 11,
+    "ା": 5,
+    "ି": 7,
+    "ୁ": 18,
+    "େ": 8,
+    "୍": 6,
+    "ୟ": 34,
+    "‍": 31
+  },
+  "dtp": {
+    "'": 14,
+    "-": 24,
+    "0": 28,
+    "1": 31,
+    "2": 33,
+    "3": 34,
+    "4": 29,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 22,
+    "c": 30,
+    "d": 9,
+    "e": 25,
+    "g": 16,
+    "h": 21,
+    "i": 6,
+    "j": 26,
+    "k": 12,
+    "l": 19,
+    "m": 13,
+    "n": 8,
+    "o": 5,
+    "p": 17,
+    "r": 18,
+    "s": 15,
+    "t": 11,
+    "u": 10,
+    "w": 23,
+    "y": 20,
+    "|": 4,
+    "—": 27
+  },
+  "dts": {
+    "'": 29,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 22,
+    "e": 11,
+    "g": 12,
+    "h": 28,
+    "i": 6,
+    "j": 23,
+    "k": 20,
+    "l": 18,
+    "m": 16,
+    "n": 8,
+    "o": 15,
+    "p": 25,
+    "r": 17,
+    "s": 21,
+    "t": 24,
+    "u": 9,
+    "w": 14,
+    "y": 19,
+    "|": 4,
+    "ŋ": 26,
+    "ɔ": 10,
+    "ɛ": 7,
+    "ɲ": 27
+  },
+  "dug": {
+    "'": 23,
+    "-": 31,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 24,
+    "d": 18,
+    "e": 10,
+    "f": 30,
+    "g": 19,
+    "h": 13,
+    "i": 6,
+    "j": 28,
+    "k": 12,
+    "l": 15,
+    "m": 9,
+    "n": 8,
+    "o": 11,
+    "p": 26,
+    "r": 16,
+    "s": 21,
+    "t": 20,
+    "u": 7,
+    "v": 27,
+    "w": 14,
+    "y": 17,
+    "z": 22,
+    "|": 5,
+    "á": 29
+  },
+  "dwr": {
+    "'": 30,
+    "-": 32,
+    "7": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 27,
+    "d": 10,
+    "e": 6,
+    "f": 29,
+    "g": 18,
+    "h": 15,
+    "i": 7,
+    "j": 31,
+    "k": 17,
+    "l": 22,
+    "m": 19,
+    "n": 8,
+    "o": 11,
+    "p": 20,
+    "q": 24,
+    "r": 23,
+    "s": 12,
+    "t": 9,
+    "u": 14,
+    "w": 16,
+    "x": 26,
+    "y": 13,
+    "z": 28,
+    "|": 5
+  },
+  "dyi": {
+    "'": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 33,
+    "d": 30,
+    "e": 8,
+    "f": 25,
+    "g": 13,
+    "h": 29,
+    "i": 7,
+    "j": 26,
+    "k": 14,
+    "l": 11,
+    "m": 19,
+    "n": 6,
+    "o": 15,
+    "p": 16,
+    "r": 18,
+    "s": 22,
+    "t": 21,
+    "u": 20,
+    "v": 35,
+    "w": 12,
+    "y": 10,
+    "z": 28,
+    "|": 4,
+    "à": 32,
+    "è": 31,
+    "ì": 27,
+    "ò": 37,
+    "ù": 36,
+    "ŋ": 23,
+    "ɔ": 17,
+    "ɛ": 9,
+    "̀": 38
+  },
+  "dyo": {
+    "'": 31,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "c": 29,
+    "d": 18,
+    "e": 10,
+    "f": 21,
+    "g": 27,
+    "h": 33,
+    "i": 6,
+    "j": 16,
+    "k": 11,
+    "l": 15,
+    "m": 9,
+    "n": 8,
+    "o": 12,
+    "p": 23,
+    "r": 17,
+    "s": 20,
+    "t": 13,
+    "u": 7,
+    "w": 25,
+    "y": 19,
+    "|": 5,
+    "á": 22,
+    "é": 32,
+    "í": 30,
+    "ñ": 28,
+    "ó": 34,
+    "ú": 24,
+    "ŋ": 26,
+    "—": 36
+  },
+  "dyu": {
+    "'": 25,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 28,
+    "d": 24,
+    "e": 11,
+    "f": 23,
+    "g": 21,
+    "h": 30,
+    "i": 7,
+    "j": 26,
+    "k": 8,
+    "l": 18,
+    "m": 15,
+    "n": 6,
+    "o": 12,
+    "p": 31,
+    "r": 13,
+    "s": 20,
+    "t": 19,
+    "u": 14,
+    "v": 32,
+    "w": 22,
+    "y": 17,
+    "z": 29,
+    "|": 4,
+    "ŋ": 33,
+    "ɔ": 10,
+    "ɛ": 9,
+    "ɲ": 27
+  },
+  "dzo": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "་": 4,
+    "ཀ": 23,
+    "ཁ": 20,
+    "ག": 7,
+    "ང": 14,
+    "ཅ": 22,
+    "ཆ": 30,
+    "ཇ": 51,
+    "ཉ": 43,
+    "ཌ": 59,
+    "ཏ": 29,
+    "ཐ": 34,
+    "ད": 8,
+    "ན": 13,
+    "པ": 21,
+    "ཕ": 37,
+    "བ": 12,
+    "མ": 15,
+    "ཙ": 45,
+    "ཚ": 25,
+    "ཛ": 46,
+    "ཝ": 39,
+    "ཞ": 35,
+    "ཟ": 26,
+    "འ": 18,
+    "ཡ": 24,
+    "ར": 17,
+    "ལ": 16,
+    "ཤ": 31,
+    "ས": 6,
+    "ཧ": 47,
+    "ཨ": 28,
+    "ཪ": 60,
+    "ཱ": 38,
+    "ི": 5,
+    "ུ": 10,
+    "ེ": 11,
+    "ོ": 9,
+    "ྐ": 36,
+    "ྒ": 40,
+    "ྔ": 53,
+    "ྕ": 57,
+    "ྗ": 52,
+    "ྙ": 55,
+    "ྟ": 33,
+    "ྡ": 42,
+    "ྣ": 54,
+    "ྤ": 48,
+    "ྦ": 32,
+    "ྨ": 44,
+    "ྩ": 49,
+    "ྫ": 56,
+    "ྭ": 58,
+    "ྱ": 19,
+    "ྲ": 27,
+    "ླ": 41,
+    "ྷ": 50
+  },
+  "eip": {
+    "'": 39,
+    "-": 26,
+    "0": 27,
+    "1": 28,
+    "2": 30,
+    "3": 36,
+    "4": 32,
+    "5": 33,
+    "6": 37,
+    "7": 34,
+    "8": 38,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 23,
+    "d": 20,
+    "e": 7,
+    "f": 25,
+    "g": 21,
+    "h": 24,
+    "i": 8,
+    "j": 29,
+    "k": 10,
+    "l": 12,
+    "m": 13,
+    "n": 6,
+    "o": 18,
+    "p": 17,
+    "r": 15,
+    "s": 19,
+    "t": 14,
+    "u": 9,
+    "w": 22,
+    "x": 40,
+    "y": 11,
+    "z": 31,
+    "|": 4
+  },
+  "eka": {
+    "'": 25,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 10,
+    "d": 27,
+    "e": 5,
+    "f": 24,
+    "g": 12,
+    "h": 14,
+    "i": 9,
+    "j": 16,
+    "k": 11,
+    "l": 13,
+    "m": 15,
+    "n": 6,
+    "o": 8,
+    "p": 23,
+    "r": 17,
+    "s": 19,
+    "t": 18,
+    "u": 20,
+    "v": 29,
+    "w": 21,
+    "y": 22,
+    "|": 4,
+    "á": 28,
+    "ń": 30,
+    "ḿ": 31
+  },
+  "ell": {
+    "!": 92,
+    "$": 98,
+    "%": 81,
+    "&": 96,
+    "'": 76,
+    "+": 99,
+    ",": 74,
+    "-": 68,
+    ".": 47,
+    "/": 77,
+    "0": 36,
+    "1": 38,
+    "2": 45,
+    "3": 58,
+    "4": 57,
+    "5": 54,
+    "6": 59,
+    "7": 69,
+    "8": 61,
+    "9": 52,
+    ":": 73,
+    ";": 93,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 88,
+    "]": 90,
+    "a": 37,
+    "b": 66,
+    "c": 51,
+    "d": 60,
+    "e": 40,
+    "f": 75,
+    "g": 65,
+    "h": 56,
+    "i": 43,
+    "j": 79,
+    "k": 67,
+    "l": 53,
+    "m": 55,
+    "n": 44,
+    "o": 41,
+    "p": 63,
+    "q": 83,
+    "r": 42,
+    "s": 46,
+    "t": 48,
+    "u": 62,
+    "v": 71,
+    "w": 72,
+    "x": 80,
+    "y": 70,
+    "z": 78,
+    "|": 4,
+    "«": 49,
+    "°": 97,
+    "²": 87,
+    "·": 86,
+    "»": 50,
+    "â": 101,
+    "é": 95,
+    "í": 91,
+    "õ": 89,
+    "́": 100,
+    "΄": 84,
+    "ΐ": 82,
+    "ά": 22,
+    "έ": 23,
+    "ή": 26,
+    "ί": 20,
+    "α": 5,
+    "β": 33,
+    "γ": 24,
+    "δ": 25,
+    "ε": 9,
+    "ζ": 35,
+    "η": 15,
+    "θ": 30,
+    "ι": 8,
+    "κ": 14,
+    "λ": 19,
+    "μ": 17,
+    "ν": 10,
+    "ξ": 34,
+    "ο": 7,
+    "π": 13,
+    "ρ": 12,
+    "ς": 16,
+    "σ": 11,
+    "τ": 6,
+    "υ": 18,
+    "φ": 31,
+    "χ": 28,
+    "ψ": 39,
+    "ω": 27,
+    "ϊ": 64,
+    "ϋ": 85,
+    "ό": 21,
+    "ύ": 29,
+    "ώ": 32,
+    "–": 94
+  },
+  "emp": {
+    "-": 59,
+    "0": 48,
+    "1": 50,
+    "2": 49,
+    "3": 55,
+    "4": 46,
+    "5": 52,
+    "6": 51,
+    "7": 57,
+    "8": 56,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 8,
+    "c": 16,
+    "d": 11,
+    "e": 9,
+    "g": 34,
+    "h": 35,
+    "i": 7,
+    "j": 12,
+    "k": 58,
+    "l": 36,
+    "m": 17,
+    "n": 18,
+    "o": 19,
+    "p": 25,
+    "q": 29,
+    "r": 6,
+    "s": 21,
+    "t": 27,
+    "u": 15,
+    "v": 40,
+    "w": 23,
+    "x": 53,
+    "y": 31,
+    "z": 33,
+    "|": 5,
+    "á": 42,
+    "â": 10,
+    "ã": 32,
+    "ä": 41,
+    "å": 22,
+    "è": 38,
+    "é": 26,
+    "ë": 45,
+    "ì": 28,
+    "ï": 30,
+    "ð": 43,
+    "ñ": 54,
+    "ò": 39,
+    "ó": 24,
+    "ô": 14,
+    "õ": 20,
+    "ö": 13,
+    "ú": 44,
+    "–": 37
+  },
+  "enb": {
+    "'": 28,
+    "-": 27,
+    "0": 29,
+    "1": 30,
+    "2": 31,
+    "4": 33,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "c": 16,
+    "e": 19,
+    "g": 25,
+    "h": 17,
+    "i": 5,
+    "k": 6,
+    "l": 22,
+    "m": 20,
+    "n": 10,
+    "o": 7,
+    "p": 18,
+    "r": 15,
+    "s": 24,
+    "t": 14,
+    "u": 11,
+    "w": 23,
+    "y": 12,
+    "|": 4,
+    "ā": 21,
+    "ē": 8,
+    "ō": 13,
+    "ʼ": 26
+  },
+  "eng": {
+    "!": 75,
+    "\"": 62,
+    "$": 48,
+    "%": 47,
+    "&": 57,
+    "'": 31,
+    "+": 56,
+    ",": 44,
+    "-": 32,
+    ".": 37,
+    "/": 45,
+    "0": 27,
+    "1": 30,
+    "2": 35,
+    "3": 41,
+    "4": 39,
+    "5": 38,
+    "6": 40,
+    "7": 43,
+    "8": 42,
+    "9": 36,
+    ":": 46,
+    ";": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 59,
+    "]": 60,
+    "`": 92,
+    "a": 7,
+    "b": 24,
+    "c": 16,
+    "d": 15,
+    "e": 5,
+    "f": 19,
+    "g": 21,
+    "h": 13,
+    "i": 9,
+    "j": 29,
+    "k": 26,
+    "l": 14,
+    "m": 18,
+    "n": 10,
+    "o": 8,
+    "p": 20,
+    "q": 34,
+    "r": 12,
+    "s": 11,
+    "t": 6,
+    "u": 17,
+    "v": 25,
+    "w": 23,
+    "x": 28,
+    "y": 22,
+    "z": 33,
+    "|": 4,
+    "£": 68,
+    "¥": 53,
+    "§": 139,
+    "°": 58,
+    "²": 76,
+    "ß": 82,
+    "à": 84,
+    "á": 49,
+    "â": 81,
+    "ã": 66,
+    "ä": 78,
+    "å": 123,
+    "æ": 110,
+    "ç": 64,
+    "è": 79,
+    "é": 55,
+    "ê": 88,
+    "ë": 87,
+    "í": 50,
+    "î": 97,
+    "ï": 85,
+    "ð": 98,
+    "ñ": 83,
+    "ò": 105,
+    "ó": 67,
+    "ô": 86,
+    "õ": 69,
+    "ö": 74,
+    "ø": 91,
+    "ú": 70,
+    "û": 99,
+    "ü": 54,
+    "ý": 112,
+    "þ": 132,
+    "ā": 100,
+    "ă": 102,
+    "ć": 94,
+    "č": 89,
+    "ē": 143,
+    "ę": 119,
+    "ě": 140,
+    "ğ": 153,
+    "ī": 107,
+    "ı": 103,
+    "ł": 73,
+    "ń": 118,
+    "ň": 137,
+    "ō": 80,
+    "ő": 120,
+    "œ": 96,
+    "ř": 108,
+    "ş": 116,
+    "š": 90,
+    "ū": 101,
+    "ž": 95,
+    "ș": 104,
+    "ə": 138,
+    "ʻ": 115,
+    "́": 93,
+    "̇": 77,
+    "α": 106,
+    "κ": 109,
+    "π": 117,
+    "χ": 141,
+    "а": 144,
+    "в": 124,
+    "е": 125,
+    "з": 130,
+    "и": 113,
+    "й": 127,
+    "к": 114,
+    "л": 126,
+    "н": 128,
+    "ь": 131,
+    "я": 129,
+    "נ": 147,
+    "ע": 148,
+    "ṃ": 121,
+    "ạ": 111,
+    "ả": 145,
+    "ị": 146,
+    "ụ": 122,
+    "​": 71,
+    "–": 61,
+    "—": 51,
+    "‘": 72,
+    "’": 65,
+    "”": 63,
+    "≡": 142,
+    "京": 149,
+    "先": 135,
+    "大": 151,
+    "尚": 134,
+    "时": 133,
+    "生": 136,
+    "都": 150,
+    "阪": 152
+  },
+  "enx": {
+    "'": 17,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 34,
+    "c": 23,
+    "d": 27,
+    "e": 6,
+    "f": 37,
+    "g": 21,
+    "h": 14,
+    "i": 26,
+    "j": 30,
+    "k": 7,
+    "l": 20,
+    "m": 9,
+    "n": 8,
+    "o": 11,
+    "p": 16,
+    "q": 25,
+    "r": 29,
+    "s": 10,
+    "t": 13,
+    "u": 31,
+    "v": 38,
+    "w": 22,
+    "x": 12,
+    "y": 15,
+    "z": 39,
+    "|": 5,
+    "à": 40,
+    "á": 19,
+    "è": 41,
+    "é": 18,
+    "í": 33,
+    "ñ": 28,
+    "ó": 24,
+    "ú": 32,
+    "ẽ": 35,
+    "—": 36
+  },
+  "epo": {
+    "'": 34,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 45,
+    "a": 5,
+    "b": 22,
+    "c": 25,
+    "d": 17,
+    "e": 8,
+    "f": 23,
+    "g": 21,
+    "h": 29,
+    "i": 6,
+    "j": 15,
+    "k": 14,
+    "l": 11,
+    "m": 18,
+    "n": 9,
+    "o": 7,
+    "p": 19,
+    "q": 46,
+    "r": 13,
+    "s": 10,
+    "t": 12,
+    "u": 16,
+    "v": 20,
+    "x": 42,
+    "y": 38,
+    "z": 26,
+    "|": 4,
+    "á": 40,
+    "é": 43,
+    "ñ": 50,
+    "ü": 47,
+    "ĉ": 28,
+    "ĝ": 24,
+    "ĥ": 33,
+    "ĵ": 32,
+    "ŝ": 30,
+    "ŭ": 27,
+    "ǔ": 37,
+    "ʼ": 49,
+    "̈": 44,
+    "̧": 48,
+    "‐": 41,
+    "–": 35,
+    "—": 36,
+    "♫": 39
+  },
+  "ese": {
+    "'": 20,
+    "-": 23,
+    "0": 33,
+    "1": 37,
+    "2": 38,
+    "3": 44,
+    "4": 39,
+    "5": 40,
+    "6": 43,
+    "7": 46,
+    "8": 48,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 12,
+    "d": 24,
+    "e": 7,
+    "f": 32,
+    "g": 28,
+    "h": 15,
+    "i": 8,
+    "j": 6,
+    "k": 47,
+    "l": 26,
+    "m": 13,
+    "n": 11,
+    "o": 9,
+    "p": 16,
+    "q": 17,
+    "r": 25,
+    "s": 18,
+    "t": 21,
+    "u": 10,
+    "v": 31,
+    "x": 42,
+    "y": 14,
+    "z": 35,
+    "|": 5,
+    "á": 34,
+    "é": 30,
+    "í": 29,
+    "ñ": 22,
+    "ó": 36,
+    "ú": 41,
+    "—": 27
+  },
+  "ess": {
+    "'": 33,
+    "-": 22,
+    "0": 32,
+    "1": 36,
+    "2": 38,
+    "3": 40,
+    "4": 37,
+    "5": 39,
+    "6": 42,
+    "7": 43,
+    "8": 44,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 27,
+    "d": 30,
+    "e": 10,
+    "f": 26,
+    "g": 5,
+    "h": 13,
+    "i": 6,
+    "j": 25,
+    "k": 14,
+    "l": 11,
+    "m": 16,
+    "n": 8,
+    "o": 28,
+    "p": 18,
+    "q": 15,
+    "r": 21,
+    "s": 17,
+    "t": 12,
+    "u": 9,
+    "v": 23,
+    "w": 20,
+    "x": 34,
+    "y": 19,
+    "z": 29,
+    "|": 7,
+    "ʼ": 45,
+    "‐": 24,
+    "ꞌ": 35
+  },
+  "est": {
+    "!": 58,
+    "\"": 36,
+    "%": 55,
+    "'": 44,
+    "+": 61,
+    ",": 51,
+    "-": 32,
+    ".": 39,
+    "/": 49,
+    "0": 29,
+    "1": 30,
+    "2": 33,
+    "3": 41,
+    "4": 40,
+    "5": 38,
+    "6": 43,
+    "7": 45,
+    "8": 42,
+    "9": 35,
+    ":": 54,
+    ";": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 66,
+    "]": 67,
+    "a": 5,
+    "b": 25,
+    "c": 31,
+    "d": 14,
+    "e": 6,
+    "f": 28,
+    "g": 20,
+    "h": 21,
+    "i": 7,
+    "j": 22,
+    "k": 13,
+    "l": 10,
+    "m": 16,
+    "n": 12,
+    "o": 15,
+    "p": 19,
+    "q": 53,
+    "r": 17,
+    "s": 8,
+    "t": 9,
+    "u": 11,
+    "v": 18,
+    "w": 37,
+    "x": 47,
+    "y": 34,
+    "z": 46,
+    "|": 4,
+    "¥": 65,
+    "°": 69,
+    "²": 68,
+    "ã": 71,
+    "ä": 23,
+    "ç": 62,
+    "é": 64,
+    "í": 70,
+    "ô": 73,
+    "õ": 24,
+    "ö": 27,
+    "ú": 72,
+    "ü": 26,
+    "ł": 63,
+    "š": 48,
+    "ž": 50,
+    "–": 52,
+    "—": 60,
+    "’": 57,
+    "”": 59
+  },
+  "eus": {
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 28,
+    "d": 15,
+    "e": 6,
+    "f": 23,
+    "g": 19,
+    "h": 20,
+    "i": 7,
+    "j": 24,
+    "k": 12,
+    "l": 16,
+    "m": 21,
+    "n": 9,
+    "o": 11,
+    "p": 22,
+    "q": 31,
+    "r": 8,
+    "s": 18,
+    "t": 10,
+    "u": 13,
+    "v": 27,
+    "w": 29,
+    "x": 25,
+    "y": 30,
+    "z": 14,
+    "|": 5,
+    "ñ": 32,
+    "—": 33
+  },
+  "evn": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 7,
+    "б": 23,
+    "в": 16,
+    "г": 21,
+    "д": 9,
+    "е": 24,
+    "з": 40,
+    "и": 11,
+    "й": 34,
+    "к": 13,
+    "л": 10,
+    "м": 17,
+    "н": 5,
+    "о": 27,
+    "п": 31,
+    "р": 14,
+    "с": 18,
+    "т": 12,
+    "у": 8,
+    "ф": 35,
+    "х": 26,
+    "ц": 38,
+    "ч": 20,
+    "ш": 39,
+    "ъ": 37,
+    "ы": 25,
+    "ь": 36,
+    "э": 6,
+    "ю": 33,
+    "я": 29,
+    "ё": 32,
+    "ӈ": 22,
+    "ӣ": 15,
+    "ӯ": 19,
+    "–": 30
+  },
+  "ewe": {
+    "'": 49,
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "d": 20,
+    "e": 5,
+    "f": 28,
+    "g": 22,
+    "h": 31,
+    "i": 7,
+    "k": 17,
+    "l": 10,
+    "m": 9,
+    "n": 12,
+    "o": 8,
+    "p": 25,
+    "r": 29,
+    "s": 15,
+    "t": 18,
+    "u": 13,
+    "v": 27,
+    "w": 14,
+    "x": 33,
+    "y": 19,
+    "z": 23,
+    "|": 4,
+    "à": 42,
+    "á": 48,
+    "ã": 30,
+    "è": 40,
+    "é": 41,
+    "ì": 51,
+    "í": 35,
+    "ò": 32,
+    "ó": 50,
+    "õ": 54,
+    "ù": 53,
+    "ú": 47,
+    "ĩ": 43,
+    "ŋ": 26,
+    "ũ": 46,
+    "ƒ": 24,
+    "ɔ": 11,
+    "ɖ": 21,
+    "ɛ": 38,
+    "ɣ": 39,
+    "ʋ": 37,
+    "̀": 52,
+    "́": 45,
+    "̃": 34,
+    "ẽ": 36
+  },
+  "eza": {
+    "'": 20,
+    "-": 27,
+    "6": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 30,
+    "d": 24,
+    "e": 5,
+    "f": 31,
+    "g": 25,
+    "h": 11,
+    "i": 10,
+    "j": 28,
+    "k": 14,
+    "l": 13,
+    "m": 22,
+    "n": 8,
+    "o": 9,
+    "p": 32,
+    "r": 21,
+    "s": 26,
+    "t": 23,
+    "u": 7,
+    "v": 35,
+    "w": 12,
+    "y": 18,
+    "z": 29,
+    "|": 4,
+    "à": 43,
+    "á": 34,
+    "è": 46,
+    "é": 36,
+    "ì": 45,
+    "í": 37,
+    "ò": 44,
+    "ó": 38,
+    "ù": 42,
+    "ú": 39,
+    "̀": 41,
+    "́": 40,
+    "ẹ": 17,
+    "ị": 33,
+    "ọ": 19,
+    "ụ": 15
+  },
+  "fal": {
+    "'": 34,
+    "-": 36,
+    "1": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 30,
+    "d": 19,
+    "e": 8,
+    "f": 28,
+    "g": 12,
+    "h": 27,
+    "i": 6,
+    "j": 24,
+    "k": 15,
+    "l": 25,
+    "m": 11,
+    "n": 7,
+    "o": 16,
+    "p": 26,
+    "r": 10,
+    "s": 20,
+    "t": 13,
+    "u": 14,
+    "v": 40,
+    "w": 17,
+    "y": 9,
+    "|": 4,
+    "à": 18,
+    "â": 21,
+    "è": 22,
+    "ê": 37,
+    "î": 29,
+    "ò": 31,
+    "ù": 38,
+    "û": 33,
+    "ɓ": 35,
+    "ɗ": 32
+  },
+  "fao": {
+    "-": 35,
+    "1": 37,
+    "4": 39,
+    "7": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 29,
+    "d": 21,
+    "e": 11,
+    "f": 23,
+    "g": 14,
+    "h": 20,
+    "i": 5,
+    "j": 27,
+    "k": 16,
+    "l": 15,
+    "m": 13,
+    "n": 9,
+    "o": 19,
+    "p": 30,
+    "r": 7,
+    "s": 12,
+    "t": 8,
+    "u": 10,
+    "v": 18,
+    "y": 25,
+    "z": 36,
+    "|": 4,
+    "á": 24,
+    "æ": 31,
+    "í": 22,
+    "ð": 17,
+    "ó": 26,
+    "ø": 28,
+    "ú": 32,
+    "ý": 33,
+    "–": 34
+  },
+  "far": {
+    "'": 18,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 26,
+    "d": 19,
+    "e": 9,
+    "f": 15,
+    "g": 20,
+    "h": 24,
+    "i": 6,
+    "j": 23,
+    "k": 10,
+    "l": 12,
+    "m": 13,
+    "n": 7,
+    "o": 11,
+    "p": 25,
+    "r": 16,
+    "s": 14,
+    "t": 17,
+    "u": 8,
+    "v": 27,
+    "w": 21,
+    "y": 28,
+    "|": 5
+  },
+  "fas": {
+    "%": 97,
+    "&": 108,
+    "'": 117,
+    "+": 105,
+    ",": 73,
+    "-": 59,
+    ".": 56,
+    "/": 79,
+    "0": 45,
+    "1": 48,
+    "2": 60,
+    "3": 61,
+    "4": 55,
+    "5": 58,
+    "6": 67,
+    "7": 83,
+    "8": 64,
+    "9": 66,
+    ":": 75,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 57,
+    "b": 86,
+    "c": 91,
+    "d": 77,
+    "e": 65,
+    "f": 99,
+    "g": 90,
+    "h": 81,
+    "i": 69,
+    "j": 104,
+    "k": 89,
+    "l": 76,
+    "m": 85,
+    "n": 74,
+    "o": 72,
+    "p": 84,
+    "q": 101,
+    "r": 70,
+    "s": 68,
+    "t": 78,
+    "u": 88,
+    "v": 92,
+    "w": 93,
+    "x": 94,
+    "y": 87,
+    "z": 109,
+    "|": 4,
+    "«": 62,
+    "»": 63,
+    "õ": 103,
+    "،": 44,
+    "؛": 96,
+    "ء": 71,
+    "آ": 29,
+    "أ": 50,
+    "ؤ": 82,
+    "ئ": 41,
+    "ا": 5,
+    "ب": 14,
+    "ة": 114,
+    "ت": 12,
+    "ث": 38,
+    "ج": 26,
+    "ح": 27,
+    "خ": 24,
+    "د": 8,
+    "ذ": 37,
+    "ر": 7,
+    "ز": 19,
+    "س": 15,
+    "ش": 18,
+    "ص": 31,
+    "ض": 34,
+    "ط": 30,
+    "ظ": 36,
+    "ع": 25,
+    "غ": 33,
+    "ف": 22,
+    "ق": 23,
+    "ك": 111,
+    "ل": 17,
+    "م": 13,
+    "ن": 9,
+    "ه": 10,
+    "و": 11,
+    "ى": 112,
+    "ي": 110,
+    "ً": 40,
+    "َ": 98,
+    "ُ": 106,
+    "ِ": 80,
+    "ّ": 95,
+    "ٔ": 113,
+    "٪": 100,
+    "پ": 28,
+    "چ": 32,
+    "ژ": 42,
+    "ک": 16,
+    "گ": 21,
+    "ۀ": 107,
+    "ی": 6,
+    "ے": 116,
+    "۰": 35,
+    "۱": 39,
+    "۲": 43,
+    "۳": 49,
+    "۴": 51,
+    "۵": 47,
+    "۶": 52,
+    "۷": 54,
+    "۸": 53,
+    "۹": 46,
+    "‌": 20,
+    "‎": 102,
+    "–": 115
+  },
+  "fij": {
+    "'": 27,
+    "-": 31,
+    "0": 32,
+    "1": 34,
+    "2": 33,
+    "3": 36,
+    "4": 38,
+    "5": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 21,
+    "d": 18,
+    "e": 9,
+    "f": 28,
+    "g": 19,
+    "i": 6,
+    "j": 26,
+    "k": 7,
+    "l": 15,
+    "m": 17,
+    "n": 8,
+    "o": 11,
+    "p": 25,
+    "q": 22,
+    "r": 16,
+    "s": 14,
+    "t": 13,
+    "u": 10,
+    "v": 12,
+    "w": 24,
+    "y": 20,
+    "|": 4,
+    "ʻ": 29,
+    "ʼ": 30,
+    "—": 35
+  },
+  "fin": {
+    "\"": 60,
+    "%": 66,
+    "'": 49,
+    "+": 68,
+    ",": 47,
+    "-": 28,
+    ".": 38,
+    "/": 50,
+    "0": 26,
+    "1": 30,
+    "2": 32,
+    "3": 39,
+    "4": 36,
+    "5": 35,
+    "6": 37,
+    "7": 41,
+    "8": 40,
+    "9": 33,
+    ":": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 61,
+    "]": 62,
+    "a": 5,
+    "b": 27,
+    "c": 29,
+    "d": 23,
+    "e": 9,
+    "f": 31,
+    "g": 25,
+    "h": 21,
+    "i": 6,
+    "j": 19,
+    "k": 13,
+    "l": 11,
+    "m": 16,
+    "n": 8,
+    "o": 12,
+    "p": 20,
+    "q": 45,
+    "r": 17,
+    "s": 10,
+    "t": 7,
+    "u": 14,
+    "v": 18,
+    "w": 34,
+    "x": 44,
+    "y": 22,
+    "z": 42,
+    "|": 4,
+    "×": 54,
+    "á": 69,
+    "ã": 64,
+    "ä": 15,
+    "ç": 56,
+    "é": 59,
+    "ë": 67,
+    "í": 63,
+    "ó": 70,
+    "õ": 57,
+    "ö": 24,
+    "ú": 58,
+    "ü": 53,
+    "ł": 55,
+    "š": 51,
+    "ž": 52,
+    "̇": 65,
+    "‎": 46,
+    "–": 48
+  },
+  "flr": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "d": 22,
+    "e": 10,
+    "f": 29,
+    "g": 15,
+    "h": 19,
+    "i": 6,
+    "j": 25,
+    "k": 9,
+    "l": 18,
+    "m": 12,
+    "n": 8,
+    "o": 13,
+    "p": 32,
+    "r": 16,
+    "s": 21,
+    "t": 20,
+    "u": 7,
+    "v": 28,
+    "w": 17,
+    "y": 14,
+    "z": 23,
+    "|": 5,
+    "á": 33,
+    "â": 27,
+    "í": 31,
+    "ù": 30,
+    "ú": 26,
+    "ꞌ": 24
+  },
+  "fmu": {
+    "'": 43,
+    "-": 37,
+    "1": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 23,
+    "अ": 20,
+    "आ": 29,
+    "इ": 30,
+    "उ": 35,
+    "ऊ": 51,
+    "ए": 38,
+    "ओ": 31,
+    "क": 9,
+    "ख": 47,
+    "ग": 25,
+    "घ": 46,
+    "च": 32,
+    "ज": 28,
+    "ट": 27,
+    "ड": 34,
+    "ण": 49,
+    "त": 12,
+    "द": 19,
+    "ध": 52,
+    "न": 6,
+    "प": 24,
+    "ब": 26,
+    "भ": 39,
+    "म": 15,
+    "य": 16,
+    "र": 7,
+    "ल": 17,
+    "ळ": 33,
+    "व": 22,
+    "श": 41,
+    "ष": 44,
+    "स": 14,
+    "ह": 21,
+    "़": 48,
+    "ा": 5,
+    "ि": 8,
+    "ी": 36,
+    "ु": 10,
+    "ू": 42,
+    "े": 11,
+    "ै": 50,
+    "ो": 13,
+    "्": 18,
+    "‍": 40
+  },
+  "fon": {
+    "'": 44,
+    "-": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 15,
+    "c": 42,
+    "d": 29,
+    "e": 16,
+    "f": 41,
+    "g": 28,
+    "h": 33,
+    "i": 20,
+    "j": 30,
+    "k": 17,
+    "l": 14,
+    "m": 10,
+    "n": 5,
+    "o": 25,
+    "p": 31,
+    "s": 26,
+    "t": 24,
+    "u": 18,
+    "v": 40,
+    "w": 13,
+    "x": 32,
+    "y": 19,
+    "z": 37,
+    "|": 4,
+    "á": 27,
+    "è": 43,
+    "é": 22,
+    "ì": 45,
+    "í": 21,
+    "ó": 12,
+    "ú": 23,
+    "đ": 47,
+    "ě": 36,
+    "ǎ": 38,
+    "ǐ": 34,
+    "ǒ": 46,
+    "ǔ": 35,
+    "ɔ": 6,
+    "ɖ": 11,
+    "ɛ": 7,
+    "́": 8,
+    "̌": 39
+  },
+  "fra": {
+    "!": 65,
+    "\"": 59,
+    "$": 80,
+    "%": 63,
+    "'": 23,
+    "+": 78,
+    ",": 61,
+    "-": 36,
+    ".": 48,
+    "/": 60,
+    "0": 32,
+    "1": 37,
+    "2": 38,
+    "3": 47,
+    "4": 46,
+    "5": 45,
+    "6": 49,
+    "7": 53,
+    "8": 50,
+    "9": 43,
+    ":": 57,
+    ";": 64,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 72,
+    "[": 74,
+    "]": 75,
+    "`": 119,
+    "a": 7,
+    "b": 25,
+    "c": 16,
+    "d": 15,
+    "e": 5,
+    "f": 22,
+    "g": 21,
+    "h": 26,
+    "i": 10,
+    "j": 31,
+    "k": 34,
+    "l": 13,
+    "m": 19,
+    "n": 8,
+    "o": 14,
+    "p": 17,
+    "q": 24,
+    "r": 11,
+    "s": 6,
+    "t": 9,
+    "u": 12,
+    "v": 20,
+    "w": 42,
+    "x": 28,
+    "y": 30,
+    "z": 33,
+    "|": 4,
+    "§": 240,
+    "«": 40,
+    "®": 184,
+    "°": 66,
+    "±": 168,
+    "²": 83,
+    "»": 41,
+    "ß": 101,
+    "à": 27,
+    "á": 69,
+    "â": 54,
+    "ã": 79,
+    "ä": 88,
+    "å": 99,
+    "æ": 85,
+    "ç": 51,
+    "è": 29,
+    "é": 18,
+    "ê": 35,
+    "ë": 73,
+    "ì": 127,
+    "í": 68,
+    "î": 52,
+    "ï": 58,
+    "ð": 108,
+    "ñ": 92,
+    "ò": 114,
+    "ó": 82,
+    "ô": 44,
+    "õ": 71,
+    "ö": 84,
+    "ø": 94,
+    "ù": 55,
+    "ú": 77,
+    "û": 56,
+    "ü": 67,
+    "ý": 107,
+    "þ": 131,
+    "ÿ": 86,
+    "ā": 95,
+    "ă": 98,
+    "ą": 115,
+    "ć": 90,
+    "ċ": 191,
+    "č": 91,
+    "ď": 187,
+    "đ": 112,
+    "ē": 124,
+    "ė": 118,
+    "ę": 111,
+    "ě": 116,
+    "ğ": 120,
+    "ġ": 289,
+    "ħ": 267,
+    "ĩ": 146,
+    "ī": 113,
+    "ı": 104,
+    "ķ": 219,
+    "ĺ": 242,
+    "ļ": 157,
+    "ľ": 160,
+    "ł": 81,
+    "ń": 96,
+    "ņ": 130,
+    "ň": 137,
+    "ō": 87,
+    "ŏ": 133,
+    "ő": 117,
+    "œ": 62,
+    "ř": 105,
+    "ś": 109,
+    "ş": 103,
+    "š": 89,
+    "ţ": 140,
+    "ť": 144,
+    "ũ": 185,
+    "ū": 93,
+    "ŭ": 221,
+    "ů": 152,
+    "ű": 209,
+    "ų": 300,
+    "ŵ": 213,
+    "ź": 126,
+    "ż": 110,
+    "ž": 97,
+    "ơ": 145,
+    "ư": 138,
+    "ǀ": 225,
+    "ǃ": 164,
+    "ǎ": 203,
+    "ǔ": 173,
+    "ǫ": 220,
+    "ǹ": 178,
+    "ș": 100,
+    "ț": 106,
+    "ɑ": 313,
+    "ə": 167,
+    "ɨ": 223,
+    "ʉ": 189,
+    "ʻ": 125,
+    "ʼ": 132,
+    "ʽ": 291,
+    "ʾ": 139,
+    "ʿ": 128,
+    "ː": 141,
+    "́": 102,
+    "̇": 121,
+    "̐": 258,
+    "̠": 275,
+    "̱": 199,
+    "̲": 312,
+    "α": 122,
+    "β": 123,
+    "γ": 143,
+    "δ": 149,
+    "ε": 165,
+    "ζ": 276,
+    "η": 290,
+    "θ": 271,
+    "ι": 250,
+    "κ": 150,
+    "λ": 153,
+    "μ": 151,
+    "ν": 154,
+    "ο": 190,
+    "π": 135,
+    "ρ": 179,
+    "ς": 252,
+    "σ": 134,
+    "τ": 188,
+    "υ": 248,
+    "φ": 196,
+    "χ": 176,
+    "ψ": 193,
+    "ω": 148,
+    "ό": 251,
+    "а": 181,
+    "г": 171,
+    "е": 180,
+    "з": 227,
+    "и": 226,
+    "к": 210,
+    "м": 228,
+    "н": 229,
+    "п": 249,
+    "р": 202,
+    "ц": 233,
+    "ч": 234,
+    "э": 292,
+    "я": 216,
+    "і": 192,
+    "ј": 166,
+    "џ": 232,
+    "ҫ": 214,
+    "ӌ": 207,
+    "գ": 277,
+    "զ": 302,
+    "ا": 170,
+    "ب": 263,
+    "ة": 262,
+    "د": 266,
+    "ر": 264,
+    "ل": 169,
+    "م": 261,
+    "ن": 197,
+    "و": 265,
+    "ي": 195,
+    "ቀ": 208,
+    "ከ": 255,
+    "ወ": 309,
+    "ደ": 198,
+    "ጀ": 274,
+    "ḍ": 217,
+    "ḥ": 158,
+    "ṅ": 218,
+    "ṇ": 162,
+    "ṣ": 129,
+    "ṭ": 155,
+    "ṯ": 186,
+    "ẓ": 269,
+    "ạ": 156,
+    "ả": 136,
+    "ầ": 161,
+    "ậ": 215,
+    "ắ": 294,
+    "ẵ": 236,
+    "ế": 222,
+    "ề": 175,
+    "ễ": 142,
+    "ệ": 147,
+    "ị": 200,
+    "ố": 260,
+    "ồ": 174,
+    "ổ": 206,
+    "ộ": 163,
+    "ớ": 293,
+    "ờ": 177,
+    "ợ": 285,
+    "ụ": 295,
+    "ủ": 224,
+    "ứ": 239,
+    "ử": 194,
+    "ỳ": 296,
+    "‐": 310,
+    "–": 70,
+    "—": 76,
+    "’": 39,
+    "′": 183,
+    "₽": 308,
+    "↔": 172,
+    "∅": 268,
+    "∆": 159,
+    "∈": 301,
+    "∞": 235,
+    "∨": 311,
+    "∼": 270,
+    "≥": 182,
+    "⊨": 247,
+    "⋅": 241,
+    "☉": 243,
+    "ⱅ": 238,
+    "ⱎ": 237,
+    "い": 259,
+    "う": 305,
+    "た": 283,
+    "つ": 284,
+    "の": 204,
+    "ひ": 205,
+    "へ": 287,
+    "ま": 288,
+    "む": 272,
+    "め": 230,
+    "も": 273,
+    "や": 231,
+    "ゔ": 306,
+    "三": 257,
+    "丹": 212,
+    "乃": 298,
+    "京": 279,
+    "保": 246,
+    "北": 278,
+    "厳": 256,
+    "宇": 244,
+    "扬": 299,
+    "文": 286,
+    "星": 201,
+    "术": 281,
+    "杜": 297,
+    "津": 245,
+    "牡": 211,
+    "甌": 254,
+    "美": 280,
+    "西": 253,
+    "貴": 304,
+    "青": 303,
+    "馆": 282,
+    "ꝑ": 307
+  },
+  "frd": {
+    "'": 31,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 30,
+    "d": 24,
+    "e": 14,
+    "f": 20,
+    "g": 23,
+    "h": 25,
+    "i": 7,
+    "j": 28,
+    "k": 16,
+    "l": 11,
+    "m": 12,
+    "n": 6,
+    "o": 13,
+    "p": 27,
+    "r": 8,
+    "s": 17,
+    "t": 10,
+    "u": 9,
+    "v": 15,
+    "w": 22,
+    "y": 19,
+    "z": 29,
+    "|": 5,
+    "ꞌ": 21
+  },
+  "fry": {
+    "'": 29,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 23,
+    "c": 26,
+    "d": 12,
+    "e": 5,
+    "f": 18,
+    "g": 22,
+    "h": 16,
+    "i": 7,
+    "j": 20,
+    "k": 15,
+    "l": 14,
+    "m": 17,
+    "n": 6,
+    "o": 13,
+    "p": 24,
+    "q": 45,
+    "r": 10,
+    "s": 11,
+    "t": 8,
+    "u": 25,
+    "v": 33,
+    "w": 19,
+    "x": 38,
+    "y": 21,
+    "z": 28,
+    "|": 4,
+    "à": 42,
+    "á": 41,
+    "â": 32,
+    "ä": 37,
+    "è": 43,
+    "é": 40,
+    "ê": 30,
+    "ë": 36,
+    "ï": 39,
+    "ô": 34,
+    "ú": 31,
+    "û": 27,
+    "ü": 44
+  },
+  "ful": {
+    "!": 58,
+    "$": 55,
+    "%": 56,
+    "&": 71,
+    "'": 35,
+    "+": 69,
+    ",": 44,
+    "-": 37,
+    ".": 33,
+    "/": 53,
+    "0": 30,
+    "1": 32,
+    "2": 36,
+    "3": 41,
+    "4": 40,
+    "5": 39,
+    "6": 42,
+    "7": 46,
+    "8": 43,
+    "9": 38,
+    ":": 49,
+    ";": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 54,
+    "[": 66,
+    "]": 67,
+    "_": 81,
+    "`": 52,
+    "a": 5,
+    "b": 14,
+    "c": 28,
+    "d": 10,
+    "e": 6,
+    "f": 24,
+    "g": 20,
+    "h": 17,
+    "i": 7,
+    "j": 19,
+    "k": 18,
+    "l": 16,
+    "m": 12,
+    "n": 9,
+    "o": 8,
+    "p": 27,
+    "q": 45,
+    "r": 15,
+    "s": 22,
+    "t": 13,
+    "u": 11,
+    "v": 29,
+    "w": 21,
+    "x": 48,
+    "y": 23,
+    "z": 34,
+    "|": 4,
+    "£": 72,
+    "¥": 75,
+    "°": 80,
+    "×": 64,
+    "á": 63,
+    "ã": 65,
+    "ç": 73,
+    "è": 76,
+    "é": 60,
+    "í": 70,
+    "ó": 78,
+    "õ": 62,
+    "ú": 82,
+    "ü": 59,
+    "ı": 68,
+    "ŋ": 77,
+    "ş": 74,
+    "ƴ": 79,
+    "ɓ": 26,
+    "ɗ": 25,
+    "ɲ": 83,
+    "‐": 84,
+    "–": 57,
+    "—": 61,
+    "‘": 50,
+    "’": 31,
+    "”": 47
+  },
+  "gag-script_cyrillic": {
+    "'": 38,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 20,
+    "в": 29,
+    "г": 26,
+    "д": 11,
+    "е": 9,
+    "ж": 37,
+    "з": 21,
+    "и": 7,
+    "й": 22,
+    "к": 13,
+    "л": 10,
+    "м": 15,
+    "н": 6,
+    "о": 16,
+    "п": 28,
+    "р": 8,
+    "с": 14,
+    "т": 19,
+    "у": 18,
+    "ф": 33,
+    "х": 24,
+    "ц": 36,
+    "ч": 27,
+    "ш": 25,
+    "ы": 12,
+    "э": 32,
+    "ӂ": 30,
+    "ӓ": 17,
+    "ӧ": 31,
+    "ӱ": 23,
+    "–": 34
+  },
+  "gag-script_latin": {
+    "'": 40,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 30,
+    "d": 11,
+    "e": 9,
+    "f": 35,
+    "g": 26,
+    "h": 24,
+    "i": 33,
+    "j": 39,
+    "k": 13,
+    "l": 10,
+    "m": 15,
+    "n": 6,
+    "o": 16,
+    "p": 28,
+    "r": 8,
+    "s": 14,
+    "t": 19,
+    "u": 18,
+    "v": 29,
+    "y": 22,
+    "z": 21,
+    "|": 4,
+    "ä": 17,
+    "ç": 27,
+    "ê": 32,
+    "ö": 31,
+    "ü": 23,
+    "ı": 12,
+    "ş": 25,
+    "ţ": 38,
+    "̇": 34,
+    "і": 7,
+    "–": 36
+  },
+  "gai": {
+    "'": 26,
+    "0": 25,
+    "1": 27,
+    "2": 28,
+    "3": 32,
+    "4": 29,
+    "5": 30,
+    "6": 34,
+    "7": 33,
+    "8": 35,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 14,
+    "d": 19,
+    "e": 12,
+    "g": 18,
+    "i": 7,
+    "k": 13,
+    "l": 23,
+    "m": 8,
+    "n": 6,
+    "o": 11,
+    "p": 17,
+    "r": 9,
+    "s": 24,
+    "t": 10,
+    "u": 16,
+    "w": 22,
+    "|": 5,
+    "ŋ": 15,
+    "ɨ": 20
+  },
+  "gam": {
+    "-": 34,
+    "0": 24,
+    "1": 25,
+    "2": 27,
+    "3": 29,
+    "4": 28,
+    "5": 30,
+    "6": 31,
+    "7": 26,
+    "8": 32,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 13,
+    "e": 6,
+    "g": 15,
+    "i": 9,
+    "j": 16,
+    "k": 12,
+    "l": 10,
+    "m": 11,
+    "n": 7,
+    "o": 8,
+    "p": 14,
+    "r": 19,
+    "s": 22,
+    "t": 23,
+    "u": 18,
+    "w": 20,
+    "y": 17,
+    "|": 4
+  },
+  "gau": {
+    "'": 61,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ం": 11,
+    "ః": 59,
+    "అ": 28,
+    "ఆ": 34,
+    "ఇ": 27,
+    "ఈ": 40,
+    "ఉ": 41,
+    "ఊ": 57,
+    "ఎ": 37,
+    "ఏ": 39,
+    "ఐ": 56,
+    "ఒ": 50,
+    "ఓ": 31,
+    "ఔ": 63,
+    "క": 12,
+    "ఖ": 55,
+    "గ": 24,
+    "ఘ": 54,
+    "చ": 26,
+    "జ": 33,
+    "ఞ": 47,
+    "ట": 20,
+    "ఠ": 60,
+    "డ": 15,
+    "ణ": 49,
+    "త": 21,
+    "థ": 58,
+    "ద": 17,
+    "ధ": 45,
+    "న": 6,
+    "ప": 16,
+    "ఫ": 52,
+    "బ": 32,
+    "భ": 44,
+    "మ": 18,
+    "య": 14,
+    "ర": 10,
+    "ల": 19,
+    "ళ": 62,
+    "వ": 22,
+    "శ": 36,
+    "ష": 43,
+    "స": 29,
+    "హ": 46,
+    "ా": 8,
+    "ి": 9,
+    "ీ": 35,
+    "ు": 7,
+    "ూ": 38,
+    "ృ": 51,
+    "ె": 13,
+    "ే": 25,
+    "ై": 42,
+    "ొ": 30,
+    "ో": 23,
+    "ౌ": 48,
+    "్": 4,
+    "‍": 53
+  },
+  "gbi": {
+    "'": 29,
+    "-": 28,
+    "0": 32,
+    "1": 33,
+    "2": 34,
+    "4": 31,
+    "5": 35,
+    "6": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 25,
+    "d": 15,
+    "e": 9,
+    "f": 27,
+    "g": 10,
+    "h": 22,
+    "i": 7,
+    "j": 26,
+    "k": 12,
+    "l": 16,
+    "m": 11,
+    "n": 8,
+    "o": 6,
+    "p": 23,
+    "r": 19,
+    "s": 13,
+    "t": 18,
+    "u": 14,
+    "w": 17,
+    "y": 21,
+    "z": 30,
+    "|": 4,
+    "ḋ": 24
+  },
+  "gbk": {
+    "'": 64,
+    "-": 58,
+    "6": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 38,
+    "ं": 26,
+    "अ": 22,
+    "आ": 41,
+    "इ": 52,
+    "ई": 36,
+    "उ": 55,
+    "ऊ": 37,
+    "ए": 60,
+    "ऐ": 42,
+    "ऑ": 66,
+    "ओ": 44,
+    "औ": 63,
+    "क": 12,
+    "ख": 43,
+    "ग": 30,
+    "घ": 57,
+    "ङ": 65,
+    "च": 33,
+    "छ": 47,
+    "ज": 17,
+    "झ": 56,
+    "ञ": 61,
+    "ट": 49,
+    "ठ": 48,
+    "ड": 35,
+    "ढ": 59,
+    "ण": 25,
+    "त": 9,
+    "थ": 34,
+    "द": 19,
+    "ध": 45,
+    "न": 13,
+    "प": 20,
+    "फ": 46,
+    "ब": 28,
+    "भ": 32,
+    "म": 15,
+    "य": 23,
+    "र": 6,
+    "ल": 24,
+    "ळ": 50,
+    "व": 31,
+    "श": 40,
+    "ष": 53,
+    "स": 11,
+    "ह": 16,
+    "़": 39,
+    "ा": 5,
+    "ि": 14,
+    "ी": 10,
+    "ु": 18,
+    "ू": 27,
+    "ृ": 62,
+    "े": 29,
+    "ै": 7,
+    "ो": 21,
+    "ौ": 54,
+    "्": 8,
+    "‍": 51,
+    "–": 68
+  },
+  "gbm": {
+    "-": 52,
+    "0": 66,
+    "5": 65,
+    "9": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 58,
+    "ं": 9,
+    "ः": 62,
+    "अ": 28,
+    "आ": 43,
+    "इ": 44,
+    "ई": 50,
+    "उ": 41,
+    "ऊ": 42,
+    "ए": 47,
+    "ऐ": 56,
+    "ओ": 60,
+    "औ": 57,
+    "क": 7,
+    "ख": 36,
+    "ग": 32,
+    "घ": 55,
+    "च": 30,
+    "छ": 34,
+    "ज": 26,
+    "झ": 54,
+    "ञ": 59,
+    "ट": 33,
+    "ठ": 49,
+    "ड": 38,
+    "ढ": 53,
+    "ण": 31,
+    "त": 13,
+    "थ": 48,
+    "द": 21,
+    "ध": 45,
+    "न": 16,
+    "प": 18,
+    "फ": 46,
+    "ब": 22,
+    "भ": 35,
+    "म": 10,
+    "य": 25,
+    "र": 6,
+    "ल": 15,
+    "ळ": 63,
+    "व": 19,
+    "श": 29,
+    "ष": 51,
+    "स": 23,
+    "ह": 27,
+    "़": 39,
+    "ा": 5,
+    "ि": 8,
+    "ी": 14,
+    "ु": 11,
+    "ू": 37,
+    "ृ": 61,
+    "े": 20,
+    "ै": 17,
+    "ो": 24,
+    "ौ": 40,
+    "्": 12,
+    "‍": 64
+  },
+  "gbo": {
+    "'": 42,
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 15,
+    "c": 38,
+    "d": 25,
+    "e": 6,
+    "f": 40,
+    "g": 41,
+    "h": 36,
+    "i": 21,
+    "j": 19,
+    "k": 11,
+    "l": 13,
+    "m": 12,
+    "n": 5,
+    "o": 9,
+    "p": 23,
+    "s": 20,
+    "t": 18,
+    "u": 14,
+    "w": 22,
+    "y": 26,
+    "|": 4,
+    "à": 29,
+    "ä": 39,
+    "è": 28,
+    "ì": 37,
+    "ï": 44,
+    "ò": 30,
+    "ù": 31,
+    "ü": 43,
+    "č": 46,
+    "ě": 17,
+    "ǎ": 32,
+    "ǐ": 33,
+    "ǒ": 16,
+    "ǔ": 34,
+    "ɔ": 10,
+    "ɕ": 47,
+    "ɛ": 7,
+    "̀": 24,
+    "̈": 35,
+    "̌": 27
+  },
+  "gde": {
+    "'": 22,
+    "-": 34,
+    "1": 35,
+    "2": 36,
+    "4": 39,
+    "5": 38,
+    "7": 40,
+    "8": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 20,
+    "d": 15,
+    "e": 24,
+    "f": 31,
+    "g": 14,
+    "h": 18,
+    "i": 7,
+    "j": 29,
+    "k": 9,
+    "l": 23,
+    "m": 12,
+    "n": 8,
+    "o": 19,
+    "p": 26,
+    "r": 25,
+    "s": 13,
+    "t": 10,
+    "u": 11,
+    "v": 32,
+    "w": 21,
+    "y": 17,
+    "z": 30,
+    "|": 4,
+    "ŋ": 16,
+    "ɓ": 33,
+    "ɗ": 28,
+    "ə": 6
+  },
+  "geb": {
+    "'": 30,
+    "0": 28,
+    "1": 29,
+    "2": 31,
+    "3": 33,
+    "4": 34,
+    "5": 32,
+    "8": 36,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 21,
+    "e": 12,
+    "f": 25,
+    "g": 8,
+    "h": 13,
+    "i": 6,
+    "k": 16,
+    "m": 9,
+    "n": 7,
+    "o": 24,
+    "p": 23,
+    "r": 11,
+    "s": 22,
+    "t": 20,
+    "u": 10,
+    "v": 17,
+    "w": 26,
+    "z": 19,
+    "|": 4,
+    "à": 15,
+    "å": 27,
+    "ç": 18
+  },
+  "gej": {
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 42,
+    "d": 34,
+    "e": 6,
+    "f": 37,
+    "g": 26,
+    "h": 39,
+    "i": 18,
+    "j": 29,
+    "k": 14,
+    "l": 19,
+    "m": 11,
+    "n": 10,
+    "o": 7,
+    "p": 35,
+    "r": 38,
+    "s": 24,
+    "t": 23,
+    "u": 13,
+    "v": 33,
+    "w": 17,
+    "x": 40,
+    "y": 21,
+    "z": 41,
+    "|": 4,
+    "à": 20,
+    "ã": 36,
+    "è": 8,
+    "ì": 28,
+    "í": 47,
+    "ò": 31,
+    "ù": 30,
+    "đ": 44,
+    "ĩ": 45,
+    "ŋ": 32,
+    "ũ": 43,
+    "ƒ": 46,
+    "ɔ": 9,
+    "ɖ": 25,
+    "ɛ": 15,
+    "̀": 12,
+    "̃": 27
+  },
+  "gil": {
+    "'": 17,
+    "-": 20,
+    "0": 21,
+    "2": 26,
+    "3": 23,
+    "4": 27,
+    "5": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "e": 8,
+    "g": 16,
+    "i": 6,
+    "k": 9,
+    "m": 13,
+    "n": 7,
+    "o": 10,
+    "p": 22,
+    "q": 25,
+    "r": 12,
+    "s": 19,
+    "t": 11,
+    "u": 15,
+    "w": 18,
+    "|": 5
+  },
+  "gjn": {
+    "'": 32,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 8,
+    "c": 29,
+    "d": 24,
+    "e": 7,
+    "f": 26,
+    "g": 27,
+    "h": 22,
+    "i": 13,
+    "j": 28,
+    "k": 10,
+    "l": 17,
+    "m": 12,
+    "n": 6,
+    "o": 11,
+    "p": 23,
+    "r": 18,
+    "s": 14,
+    "t": 20,
+    "u": 16,
+    "v": 31,
+    "w": 25,
+    "y": 21,
+    "z": 30,
+    "|": 4,
+    "ŋ": 15,
+    "ɔ": 19,
+    "ɛ": 9
+  },
+  "gkn": {
+    "'": 53,
+    "-": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 5,
+    "d": 21,
+    "e": 6,
+    "f": 44,
+    "g": 13,
+    "h": 50,
+    "i": 23,
+    "j": 43,
+    "k": 12,
+    "l": 9,
+    "m": 19,
+    "n": 7,
+    "o": 16,
+    "p": 31,
+    "r": 34,
+    "s": 30,
+    "t": 22,
+    "u": 33,
+    "v": 27,
+    "y": 35,
+    "z": 36,
+    "|": 4,
+    "à": 11,
+    "á": 10,
+    "ã": 38,
+    "è": 15,
+    "é": 18,
+    "ì": 28,
+    "í": 20,
+    "ò": 24,
+    "ó": 17,
+    "õ": 37,
+    "ù": 39,
+    "ú": 32,
+    "ĩ": 41,
+    "ń": 47,
+    "ũ": 54,
+    "ǹ": 56,
+    "ʻ": 55,
+    "ʽ": 48,
+    "̀": 25,
+    "́": 29,
+    "̃": 46,
+    "̄": 45,
+    "ḿ": 49,
+    "ṍ": 42,
+    "ṹ": 52,
+    "ẹ": 26,
+    "ẽ": 40,
+    "ọ": 14,
+    "ỳ": 57
+  },
+  "gld": {
+    "'": 35,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 6,
+    "б": 16,
+    "в": 23,
+    "г": 20,
+    "д": 9,
+    "е": 29,
+    "з": 33,
+    "и": 5,
+    "й": 19,
+    "к": 22,
+    "л": 15,
+    "м": 13,
+    "н": 8,
+    "о": 12,
+    "п": 26,
+    "р": 17,
+    "с": 11,
+    "т": 18,
+    "у": 10,
+    "ф": 31,
+    "х": 21,
+    "ц": 36,
+    "ч": 14,
+    "ь": 34,
+    "э": 7,
+    "ю": 30,
+    "я": 28,
+    "ё": 25,
+    "ӈ": 24,
+    "–": 32
+  },
+  "gle": {
+    "!": 73,
+    "\"": 59,
+    "$": 53,
+    "%": 57,
+    "&": 63,
+    "'": 31,
+    "+": 62,
+    ",": 49,
+    "-": 28,
+    ".": 38,
+    "/": 50,
+    "0": 29,
+    "1": 30,
+    "2": 34,
+    "3": 44,
+    "4": 42,
+    "5": 39,
+    "6": 41,
+    "7": 45,
+    "8": 43,
+    "9": 36,
+    ":": 51,
+    ";": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 75,
+    "]": 76,
+    "a": 5,
+    "b": 22,
+    "c": 13,
+    "d": 16,
+    "e": 11,
+    "f": 24,
+    "g": 17,
+    "h": 8,
+    "i": 6,
+    "j": 40,
+    "k": 32,
+    "l": 15,
+    "m": 19,
+    "n": 7,
+    "o": 14,
+    "p": 27,
+    "q": 52,
+    "r": 9,
+    "s": 12,
+    "t": 10,
+    "u": 18,
+    "v": 33,
+    "w": 37,
+    "x": 47,
+    "y": 35,
+    "z": 46,
+    "|": 4,
+    "£": 66,
+    "¥": 70,
+    "°": 68,
+    "²": 74,
+    "á": 20,
+    "ã": 72,
+    "ç": 60,
+    "é": 23,
+    "í": 21,
+    "ó": 26,
+    "õ": 61,
+    "ö": 64,
+    "ú": 25,
+    "ü": 54,
+    "ł": 77,
+    "̇": 71,
+    "​": 48,
+    "–": 69,
+    "—": 58,
+    "‘": 67,
+    "’": 55,
+    "”": 65
+  },
+  "glg": {
+    "!": 59,
+    "\"": 66,
+    "%": 55,
+    "&": 61,
+    "'": 54,
+    "+": 69,
+    ",": 53,
+    "-": 51,
+    ".": 48,
+    "/": 52,
+    "0": 33,
+    "1": 34,
+    "2": 36,
+    "3": 44,
+    "4": 43,
+    "5": 42,
+    "6": 47,
+    "7": 49,
+    "8": 46,
+    "9": 41,
+    ":": 50,
+    ";": 57,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 65,
+    "[": 72,
+    "]": 73,
+    "a": 5,
+    "b": 19,
+    "c": 14,
+    "d": 12,
+    "e": 6,
+    "f": 22,
+    "g": 21,
+    "h": 25,
+    "i": 11,
+    "j": 45,
+    "k": 35,
+    "l": 16,
+    "m": 17,
+    "n": 9,
+    "o": 7,
+    "p": 18,
+    "q": 24,
+    "r": 10,
+    "s": 8,
+    "t": 13,
+    "u": 15,
+    "v": 20,
+    "w": 38,
+    "x": 23,
+    "y": 37,
+    "z": 30,
+    "|": 4,
+    "¥": 64,
+    "ª": 62,
+    "«": 39,
+    "°": 60,
+    "²": 74,
+    "º": 58,
+    "»": 40,
+    "á": 28,
+    "ã": 70,
+    "ç": 63,
+    "é": 29,
+    "í": 27,
+    "ñ": 32,
+    "ó": 26,
+    "ö": 71,
+    "ú": 31,
+    "ü": 56,
+    "ł": 68,
+    "–": 67,
+    "—": 75
+  },
+  "glk": {
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "آ": 32,
+    "أ": 14,
+    "ؤ": 41,
+    "ئ": 33,
+    "ا": 7,
+    "ب": 13,
+    "ت": 15,
+    "ث": 38,
+    "ج": 26,
+    "ح": 30,
+    "خ": 22,
+    "د": 9,
+    "ذ": 39,
+    "ر": 10,
+    "ز": 21,
+    "س": 17,
+    "ش": 18,
+    "ص": 34,
+    "ض": 37,
+    "ط": 31,
+    "ظ": 36,
+    "ع": 25,
+    "غ": 35,
+    "ف": 24,
+    "ق": 29,
+    "ل": 23,
+    "م": 12,
+    "ن": 8,
+    "ه": 11,
+    "و": 6,
+    "ٚ": 19,
+    "پ": 27,
+    "چ": 28,
+    "ک": 16,
+    "گ": 20,
+    "ی": 5
+  },
+  "gmv": {
+    "-": 31,
+    "1": 33,
+    "2": 32,
+    "7": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 27,
+    "d": 12,
+    "e": 8,
+    "f": 28,
+    "g": 16,
+    "h": 13,
+    "i": 6,
+    "j": 30,
+    "k": 15,
+    "l": 18,
+    "m": 17,
+    "n": 11,
+    "o": 10,
+    "p": 23,
+    "q": 24,
+    "r": 21,
+    "s": 7,
+    "t": 9,
+    "u": 19,
+    "w": 25,
+    "x": 26,
+    "y": 14,
+    "z": 20,
+    "|": 5
+  },
+  "gna": {
+    "'": 26,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 17,
+    "c": 61,
+    "d": 28,
+    "e": 33,
+    "f": 39,
+    "g": 29,
+    "h": 16,
+    "i": 24,
+    "j": 44,
+    "k": 12,
+    "l": 42,
+    "m": 14,
+    "n": 9,
+    "o": 31,
+    "p": 18,
+    "r": 10,
+    "s": 20,
+    "t": 21,
+    "u": 34,
+    "v": 50,
+    "w": 19,
+    "y": 15,
+    "z": 45,
+    "|": 4,
+    "á": 11,
+    "â": 43,
+    "ã": 30,
+    "é": 23,
+    "ê": 51,
+    "í": 27,
+    "î": 52,
+    "ó": 37,
+    "ô": 53,
+    "õ": 55,
+    "ú": 40,
+    "û": 56,
+    "ě": 59,
+    "ĩ": 41,
+    "ŋ": 32,
+    "ũ": 46,
+    "ǎ": 48,
+    "ǐ": 58,
+    "ǒ": 54,
+    "ɔ": 22,
+    "ɛ": 8,
+    "ɩ": 5,
+    "ɲ": 35,
+    "ʋ": 13,
+    "́": 6,
+    "̂": 36,
+    "̃": 25,
+    "̌": 47,
+    "ṍ": 57,
+    "ṹ": 49,
+    "–": 60
+  },
+  "gnd": {
+    "'": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "d": 18,
+    "e": 10,
+    "f": 31,
+    "g": 14,
+    "h": 28,
+    "i": 12,
+    "k": 7,
+    "l": 16,
+    "m": 6,
+    "n": 19,
+    "o": 36,
+    "p": 32,
+    "r": 11,
+    "s": 17,
+    "t": 9,
+    "u": 15,
+    "v": 30,
+    "w": 21,
+    "y": 22,
+    "z": 24,
+    "|": 4,
+    "à": 8,
+    "è": 29,
+    "ì": 27,
+    "ù": 33,
+    "ŋ": 23,
+    "ɂ": 37,
+    "ɓ": 34,
+    "ɗ": 26,
+    "ə": 13,
+    "̀": 25
+  },
+  "gng": {
+    "'": 19,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 8,
+    "c": 32,
+    "d": 23,
+    "e": 10,
+    "f": 33,
+    "g": 30,
+    "h": 24,
+    "i": 6,
+    "j": 29,
+    "k": 11,
+    "l": 15,
+    "m": 17,
+    "n": 5,
+    "o": 18,
+    "p": 27,
+    "r": 21,
+    "s": 22,
+    "t": 14,
+    "u": 12,
+    "w": 20,
+    "y": 16,
+    "|": 4,
+    "à": 25,
+    "á": 42,
+    "è": 40,
+    "é": 38,
+    "ì": 35,
+    "í": 28,
+    "ñ": 36,
+    "ò": 45,
+    "ó": 41,
+    "ù": 39,
+    "ú": 46,
+    "ń": 34,
+    "ŋ": 31,
+    "ɔ": 13,
+    "ɛ": 9,
+    "ʼ": 26,
+    "̀": 37,
+    "́": 44
+  },
+  "gof-script_latin": {
+    "-": 31,
+    "7": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 27,
+    "d": 14,
+    "e": 8,
+    "f": 28,
+    "g": 17,
+    "h": 12,
+    "i": 6,
+    "j": 30,
+    "k": 15,
+    "l": 20,
+    "m": 18,
+    "n": 9,
+    "o": 10,
+    "p": 24,
+    "q": 25,
+    "r": 21,
+    "s": 7,
+    "t": 11,
+    "u": 16,
+    "w": 22,
+    "x": 26,
+    "y": 13,
+    "z": 29,
+    "|": 5
+  },
+  "gog": {
+    "'": 27,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 21,
+    "d": 25,
+    "e": 11,
+    "f": 28,
+    "g": 17,
+    "h": 16,
+    "i": 7,
+    "j": 24,
+    "k": 15,
+    "l": 12,
+    "m": 13,
+    "n": 8,
+    "o": 9,
+    "p": 26,
+    "s": 19,
+    "t": 20,
+    "u": 6,
+    "v": 23,
+    "w": 10,
+    "y": 14,
+    "z": 18,
+    "|": 4
+  },
+  "gor": {
+    "'": 22,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 19,
+    "c": 27,
+    "d": 18,
+    "e": 15,
+    "f": 29,
+    "g": 16,
+    "h": 17,
+    "i": 8,
+    "j": 26,
+    "k": 24,
+    "l": 7,
+    "m": 11,
+    "n": 13,
+    "o": 5,
+    "p": 20,
+    "r": 23,
+    "s": 21,
+    "t": 10,
+    "u": 9,
+    "w": 12,
+    "y": 14,
+    "z": 28,
+    "|": 4
+  },
+  "gqr": {
+    "'": 11,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "d": 13,
+    "e": 10,
+    "g": 15,
+    "h": 29,
+    "i": 9,
+    "j": 20,
+    "k": 8,
+    "l": 16,
+    "m": 19,
+    "n": 5,
+    "o": 17,
+    "p": 28,
+    "r": 18,
+    "s": 24,
+    "t": 12,
+    "u": 22,
+    "w": 25,
+    "y": 23,
+    "|": 4,
+    "ɓ": 27,
+    "ɔ": 21,
+    "ɗ": 30,
+    "ə": 14,
+    "ɨ": 6
+  },
+  "grc": {
+    "'": 117,
+    "-": 108,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ΐ": 99,
+    "ά": 32,
+    "έ": 27,
+    "ή": 42,
+    "ί": 24,
+    "ΰ": 116,
+    "α": 6,
+    "β": 44,
+    "γ": 21,
+    "δ": 18,
+    "ε": 9,
+    "��": 57,
+    "η": 26,
+    "θ": 23,
+    "ι": 11,
+    "κ": 13,
+    "λ": 17,
+    "μ": 16,
+    "ν": 5,
+    "ξ": 51,
+    "ο": 8,
+    "π": 15,
+    "ρ": 14,
+    "ς": 10,
+    "σ": 12,
+    "τ": 7,
+    "υ": 22,
+    "φ": 40,
+    "χ": 36,
+    "ψ": 62,
+    "ω": 30,
+    "ϊ": 97,
+    "ϋ": 96,
+    "ό": 33,
+    "ύ": 39,
+    "ώ": 53,
+    "ἀ": 29,
+    "ἁ": 71,
+    "ἂ": 88,
+    "ἃ": 94,
+    "ἄ": 60,
+    "ἅ": 82,
+    "ἆ": 113,
+    "ἐ": 19,
+    "ἑ": 65,
+    "ἓ": 98,
+    "ἔ": 49,
+    "ἕ": 76,
+    "ἠ": 81,
+    "ἡ": 50,
+    "ἢ": 79,
+    "ἣ": 92,
+    "ἤ": 80,
+    "ἥ": 89,
+    "ἦ": 74,
+    "ἧ": 101,
+    "ἰ": 34,
+    "ἱ": 54,
+    "ἳ": 105,
+    "ἴ": 69,
+    "ἵ": 66,
+    "ἶ": 59,
+    "ἷ": 84,
+    "ὀ": 70,
+    "ὁ": 47,
+    "ὃ": 72,
+    "ὄ": 73,
+    "ὅ": 55,
+    "ὐ": 28,
+    "ὑ": 46,
+    "ὒ": 119,
+    "ὓ": 103,
+    "ὔ": 86,
+    "ὕ": 77,
+    "ὖ": 78,
+    "ὗ": 75,
+    "ὠ": 100,
+    "ὡ": 68,
+    "ὢ": 109,
+    "ὤ": 102,
+    "ὥ": 83,
+    "ὦ": 107,
+    "ὧ": 87,
+    "ὰ": 38,
+    "ὲ": 43,
+    "ὴ": 41,
+    "ὶ": 20,
+    "ὸ": 31,
+    "ὺ": 52,
+    "ὼ": 64,
+    "ᾄ": 118,
+    "ᾅ": 115,
+    "ᾐ": 114,
+    "ᾑ": 121,
+    "ᾔ": 112,
+    "ᾖ": 104,
+    "ᾗ": 106,
+    "ᾠ": 111,
+    "ᾧ": 90,
+    "ᾳ": 67,
+    "ᾶ": 56,
+    "ᾷ": 91,
+    "ῃ": 63,
+    "ῄ": 110,
+    "ῆ": 45,
+    "ῇ": 58,
+    "ῒ": 95,
+    "ῖ": 37,
+    "ῥ": 85,
+    "ῦ": 25,
+    "ῳ": 61,
+    "ῴ": 93,
+    "ῶ": 35,
+    "ῷ": 48,
+    "–": 120
+  },
+  "gri": {
+    "'": 25,
+    "-": 29,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "4": 32,
+    "6": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "d": 19,
+    "e": 12,
+    "g": 11,
+    "h": 24,
+    "i": 6,
+    "j": 23,
+    "k": 16,
+    "l": 17,
+    "m": 13,
+    "n": 7,
+    "o": 8,
+    "p": 20,
+    "q": 22,
+    "r": 14,
+    "s": 15,
+    "t": 10,
+    "u": 9,
+    "v": 18,
+    "|": 5,
+    "–": 27
+  },
+  "grn": {
+    "'": 17,
+    "-": 43,
+    "0": 47,
+    "1": 48,
+    "2": 50,
+    "3": 56,
+    "4": 52,
+    "5": 51,
+    "6": 53,
+    "7": 49,
+    "8": 54,
+    "9": 57,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 26,
+    "d": 27,
+    "e": 6,
+    "f": 44,
+    "g": 22,
+    "h": 10,
+    "i": 8,
+    "j": 20,
+    "k": 14,
+    "l": 39,
+    "m": 13,
+    "n": 18,
+    "o": 7,
+    "p": 12,
+    "q": 45,
+    "r": 9,
+    "s": 28,
+    "t": 15,
+    "u": 11,
+    "v": 16,
+    "x": 55,
+    "y": 19,
+    "z": 42,
+    "|": 4,
+    "á": 24,
+    "ã": 29,
+    "é": 21,
+    "í": 35,
+    "ñ": 25,
+    "ó": 31,
+    "õ": 32,
+    "ú": 36,
+    "ý": 38,
+    "ĩ": 33,
+    "ũ": 40,
+    "ʼ": 34,
+    "̃": 37,
+    "ẽ": 30,
+    "ỹ": 41,
+    "—": 46
+  },
+  "grt": {
+    "'": 46,
+    "-": 43,
+    "1": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 65,
+    "d": 64,
+    "l": 63,
+    "s": 62,
+    "|": 4,
+    "ঁ": 61,
+    "ং": 8,
+    "ঃ": 53,
+    "অ": 42,
+    "আ": 16,
+    "ই": 28,
+    "ঈ": 60,
+    "উ": 33,
+    "ঊ": 58,
+    "এ": 36,
+    "ঐ": 56,
+    "ও": 30,
+    "ক": 10,
+    "খ": 17,
+    "গ": 18,
+    "ঘ": 52,
+    "ঙ": 57,
+    "চ": 25,
+    "ছ": 48,
+    "জ": 21,
+    "ঝ": 55,
+    "ট": 40,
+    "ঠ": 51,
+    "ড": 45,
+    "ণ": 47,
+    "ত": 26,
+    "থ": 31,
+    "দ": 19,
+    "ধ": 49,
+    "ন": 7,
+    "প": 32,
+    "ফ": 34,
+    "ব": 12,
+    "ভ": 54,
+    "ম": 11,
+    "য": 22,
+    "র": 13,
+    "ল": 24,
+    "শ": 38,
+    "ষ": 39,
+    "স": 14,
+    "হ": 29,
+    "়": 23,
+    "া": 5,
+    "ি": 6,
+    "ী": 35,
+    "ু": 27,
+    "ূ": 41,
+    "ে": 20,
+    "ৈ": 50,
+    "ো": 15,
+    "ৌ": 44,
+    "্": 9,
+    "ৎ": 37
+  },
+  "gso": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 32,
+    "d": 20,
+    "e": 12,
+    "f": 27,
+    "g": 11,
+    "h": 13,
+    "i": 8,
+    "j": 30,
+    "k": 14,
+    "l": 22,
+    "m": 9,
+    "n": 6,
+    "o": 15,
+    "p": 23,
+    "q": 35,
+    "r": 26,
+    "s": 17,
+    "t": 19,
+    "u": 21,
+    "v": 34,
+    "w": 16,
+    "x": 36,
+    "y": 25,
+    "z": 24,
+    "|": 4,
+    "è": 37,
+    "é": 31,
+    "ë": 38,
+    "ï": 33,
+    "ɔ": 10,
+    "ɛ": 7,
+    "̂": 29
+  },
+  "gub": {
+    "'": 12,
+    "-": 26,
+    "0": 24,
+    "1": 25,
+    "2": 29,
+    "3": 34,
+    "4": 28,
+    "5": 30,
+    "6": 31,
+    "7": 27,
+    "8": 32,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "e": 5,
+    "g": 22,
+    "h": 11,
+    "i": 14,
+    "k": 17,
+    "m": 15,
+    "n": 13,
+    "o": 18,
+    "p": 10,
+    "r": 9,
+    "t": 19,
+    "u": 7,
+    "w": 8,
+    "x": 23,
+    "y": 21,
+    "z": 16,
+    "|": 4,
+    "à": 20
+  },
+  "guc": {
+    "'": 17,
+    "0": 35,
+    "1": 39,
+    "2": 40,
+    "3": 45,
+    "4": 43,
+    "5": 42,
+    "6": 46,
+    "7": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 24,
+    "d": 25,
+    "e": 9,
+    "f": 32,
+    "g": 31,
+    "h": 19,
+    "i": 7,
+    "j": 14,
+    "k": 12,
+    "l": 15,
+    "m": 16,
+    "n": 6,
+    "o": 18,
+    "p": 21,
+    "q": 38,
+    "r": 23,
+    "s": 10,
+    "t": 11,
+    "u": 13,
+    "v": 36,
+    "w": 20,
+    "x": 41,
+    "y": 22,
+    "z": 37,
+    "|": 5,
+    "á": 33,
+    "é": 30,
+    "í": 29,
+    "ñ": 44,
+    "ó": 34,
+    "ú": 26,
+    "ü": 8,
+    "—": 27
+  },
+  "gud": {
+    "'": 9,
+    "0": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 22,
+    "d": 30,
+    "e": 19,
+    "f": 31,
+    "g": 21,
+    "h": 24,
+    "i": 16,
+    "j": 33,
+    "k": 14,
+    "l": 6,
+    "m": 18,
+    "n": 7,
+    "o": 23,
+    "p": 25,
+    "s": 26,
+    "t": 20,
+    "u": 27,
+    "v": 34,
+    "w": 28,
+    "y": 12,
+    "z": 29,
+    "|": 4,
+    "ŋ": 32,
+    "ɔ": 10,
+    "ɛ": 13,
+    "ɩ": 8,
+    "ʋ": 15,
+    "‐": 11
+  },
+  "guh": {
+    "'": 40,
+    "-": 23,
+    "4": 47,
+    "7": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "c": 15,
+    "d": 25,
+    "e": 6,
+    "f": 30,
+    "g": 31,
+    "h": 29,
+    "i": 5,
+    "j": 10,
+    "k": 42,
+    "l": 24,
+    "m": 20,
+    "n": 8,
+    "o": 13,
+    "p": 11,
+    "q": 33,
+    "r": 21,
+    "s": 12,
+    "t": 9,
+    "u": 16,
+    "v": 37,
+    "w": 17,
+    "x": 19,
+    "y": 22,
+    "z": 35,
+    "|": 7,
+    "­": 41,
+    "á": 34,
+    "ã": 39,
+    "é": 32,
+    "ë": 18,
+    "í": 28,
+    "ñ": 43,
+    "ó": 36,
+    "õ": 44,
+    "ú": 26,
+    "ĩ": 46,
+    "̱": 38,
+    "—": 27
+  },
+  "guj": {
+    "!": 109,
+    "\"": 50,
+    "$": 100,
+    "%": 106,
+    "&": 118,
+    "'": 81,
+    "+": 110,
+    ",": 64,
+    "-": 57,
+    ".": 54,
+    "/": 80,
+    "0": 49,
+    "1": 53,
+    "2": 56,
+    "3": 65,
+    "4": 66,
+    "5": 63,
+    "6": 69,
+    "7": 70,
+    "8": 68,
+    "9": 58,
+    ":": 75,
+    ";": 96,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 111,
+    "]": 112,
+    "a": 77,
+    "b": 95,
+    "c": 85,
+    "d": 90,
+    "e": 83,
+    "f": 98,
+    "g": 92,
+    "h": 104,
+    "i": 76,
+    "j": 113,
+    "k": 105,
+    "l": 107,
+    "m": 87,
+    "n": 82,
+    "o": 84,
+    "p": 79,
+    "q": 117,
+    "r": 93,
+    "s": 78,
+    "t": 88,
+    "u": 94,
+    "v": 97,
+    "w": 103,
+    "x": 101,
+    "y": 108,
+    "z": 116,
+    "|": 4,
+    "¥": 114,
+    "°": 115,
+    "õ": 123,
+    "ં": 17,
+    "ઃ": 91,
+    "અ": 28,
+    "આ": 34,
+    "ઇ": 47,
+    "ઈ": 46,
+    "ઉ": 48,
+    "ઊ": 72,
+    "ઋ": 99,
+    "ઍ": 86,
+    "એ": 36,
+    "ઐ": 102,
+    "ઑ": 73,
+    "ઓ": 38,
+    "ઔ": 89,
+    "ક": 10,
+    "ખ": 43,
+    "ગ": 30,
+    "ઘ": 52,
+    "ચ": 39,
+    "છ": 27,
+    "જ": 23,
+    "ઝ": 51,
+    "ઞ": 74,
+    "ટ": 25,
+    "ઠ": 59,
+    "ડ": 35,
+    "ઢ": 71,
+    "ણ": 33,
+    "ત": 12,
+    "થ": 26,
+    "દ": 32,
+    "ધ": 37,
+    "ન": 8,
+    "પ": 19,
+    "ફ": 41,
+    "બ": 31,
+    "ભ": 44,
+    "મ": 13,
+    "ય": 18,
+    "ર": 7,
+    "લ": 21,
+    "ળ": 42,
+    "વ": 14,
+    "શ": 29,
+    "ષ": 45,
+    "સ": 16,
+    "હ": 24,
+    "઼": 119,
+    "ા": 5,
+    "િ": 20,
+    "ી": 11,
+    "ુ": 22,
+    "ૂ": 40,
+    "ૃ": 55,
+    "ૅ": 67,
+    "ે": 6,
+    "ૈ": 61,
+    "ૉ": 62,
+    "ો": 15,
+    "ૌ": 60,
+    "્": 9,
+    "૪": 121,
+    "૬": 122,
+    "૮": 120
+  },
+  "guk": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "w": 220,
+    "|": 4,
+    "ä": 216,
+    "ሀ": 207,
+    "ሁ": 86,
+    "ሂ": 126,
+    "ሃ": 104,
+    "ሄ": 102,
+    "ህ": 199,
+    "ሆ": 148,
+    "ለ": 16,
+    "ሉ": 107,
+    "ሊ": 146,
+    "ላ": 37,
+    "ሌ": 84,
+    "ል": 15,
+    "ሎ": 67,
+    "መ": 12,
+    "ሙ": 72,
+    "ሚ": 103,
+    "ማ": 6,
+    "ሜ": 56,
+    "ም": 9,
+    "ሞ": 136,
+    "ሠ": 55,
+    "ሡ": 217,
+    "ሢ": 208,
+    "ሣ": 70,
+    "ሤ": 115,
+    "ሥ": 20,
+    "ሦ": 198,
+    "ረ": 44,
+    "ሩ": 128,
+    "ሪ": 133,
+    "ራ": 41,
+    "ሬ": 138,
+    "ር": 26,
+    "ሮ": 81,
+    "ሰ": 49,
+    "ሱ": 65,
+    "ሲ": 97,
+    "ሳ": 61,
+    "ሴ": 122,
+    "ስ": 25,
+    "ሶ": 118,
+    "ሸ": 40,
+    "ሹ": 158,
+    "ሺ": 181,
+    "ሻ": 54,
+    "ሼ": 184,
+    "ሽ": 50,
+    "ሾ": 87,
+    "ቀ": 124,
+    "ቁ": 135,
+    "ቂ": 173,
+    "ቃ": 161,
+    "ቄ": 176,
+    "ቅ": 175,
+    "ቆ": 22,
+    "ቊ": 194,
+    "ቋ": 76,
+    "ቌ": 189,
+    "በ": 11,
+    "ቢ": 35,
+    "ባ": 30,
+    "ቤ": 24,
+    "ብ": 71,
+    "ቦ": 90,
+    "ቧ": 197,
+    "ተ": 19,
+    "ቱ": 131,
+    "ቲ": 129,
+    "��": 53,
+    "ቴ": 121,
+    "ት": 52,
+    "ቶ": 63,
+    "ቸ": 88,
+    "ቹ": 187,
+    "ቻ": 110,
+    "ቼ": 139,
+    "ች": 151,
+    "ቾ": 95,
+    "ነ": 23,
+    "ኑ": 193,
+    "ኒ": 94,
+    "ና": 48,
+    "ኔ": 101,
+    "ን": 13,
+    "ኖ": 51,
+    "አ": 8,
+    "ኡ": 125,
+    "ኢ": 38,
+    "ኣ": 42,
+    "ኤ": 10,
+    "እ": 27,
+    "ኦ": 66,
+    "ከ": 5,
+    "ኩ": 113,
+    "ኪ": 82,
+    "ካ": 14,
+    "ኬ": 89,
+    "ክ": 79,
+    "ኮ": 31,
+    "ኲ": 191,
+    "ኳ": 17,
+    "ኴ": 201,
+    "ኸ": 32,
+    "ኹ": 169,
+    "ኺ": 159,
+    "ኻ": 36,
+    "ኼ": 39,
+    "ኽ": 43,
+    "ኾ": 145,
+    "ዉ": 111,
+    "ዊ": 154,
+    "ዋ": 68,
+    "ዌ": 96,
+    "ው": 219,
+    "ዎ": 47,
+    "ዘ": 83,
+    "ዙ": 171,
+    "ዚ": 150,
+    "ዛ": 80,
+    "ዜ": 109,
+    "ዝ": 117,
+    "ዞ": 179,
+    "ዠ": 127,
+    "ዡ": 190,
+    "ዢ": 114,
+    "ዣ": 108,
+    "ዤ": 210,
+    "ዥ": 46,
+    "ዦ": 147,
+    "የ": 218,
+    "ዩ": 178,
+    "ዪ": 45,
+    "ያ": 18,
+    "ዬ": 98,
+    "ይ": 77,
+    "ዮ": 132,
+    "ደ": 21,
+    "ዱ": 75,
+    "ዲ": 137,
+    "ዳ": 29,
+    "ዴ": 60,
+    "ድ": 73,
+    "ዶ": 58,
+    "ዸ": 116,
+    "ዹ": 185,
+    "ዺ": 203,
+    "ዻ": 112,
+    "ዼ": 202,
+    "ዽ": 143,
+    "ዾ": 162,
+    "ጀ": 99,
+    "ጂ": 168,
+    "ጃ": 100,
+    "ጅ": 120,
+    "ጆ": 206,
+    "ገ": 7,
+    "ጉ": 119,
+    "ጊ": 192,
+    "ጋ": 28,
+    "ጌ": 164,
+    "ግ": 62,
+    "ጎ": 33,
+    "ጒ": 215,
+    "ጓ": 74,
+    "ጔ": 214,
+    "ጠ": 130,
+    "ጡ": 195,
+    "ጢ": 183,
+    "ጣ": 177,
+    "ጤ": 141,
+    "ጥ": 91,
+    "ጦ": 64,
+    "ጨ": 78,
+    "ጩ": 157,
+    "ጪ": 204,
+    "ጫ": 93,
+    "ጬ": 212,
+    "ጭ": 165,
+    "ጮ": 155,
+    "ጰ": 213,
+    "ጲ": 200,
+    "ጳ": 123,
+    "ጴ": 140,
+    "ጵ": 172,
+    "ጶ": 174,
+    "ጸ": 34,
+    "ጹ": 149,
+    "ጺ": 152,
+    "ጻ": 134,
+    "ጼ": 209,
+    "ጽ": 153,
+    "ጾ": 180,
+    "ፀ": 57,
+    "ፁ": 166,
+    "ፂ": 170,
+    "ፃ": 105,
+    "ፄ": 59,
+    "ፅ": 163,
+    "ፆ": 144,
+    "ፈ": 69,
+    "ፉ": 92,
+    "ፊ": 188,
+    "ፋ": 85,
+    "ፌ": 167,
+    "ፍ": 156,
+    "ፎ": 211,
+    "ፐ": 106,
+    "ፑ": 160,
+    "ፒ": 205,
+    "ፓ": 182,
+    "ፔ": 186,
+    "ፕ": 196,
+    "ፖ": 142
+  },
+  "gum": {
+    "'": 34,
+    "-": 37,
+    "0": 40,
+    "1": 42,
+    "2": 43,
+    "3": 45,
+    "4": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 19,
+    "d": 24,
+    "e": 9,
+    "f": 35,
+    "g": 12,
+    "h": 18,
+    "i": 6,
+    "j": 28,
+    "k": 17,
+    "l": 20,
+    "m": 13,
+    "n": 8,
+    "o": 25,
+    "p": 22,
+    "q": 39,
+    "r": 7,
+    "s": 16,
+    "t": 15,
+    "u": 11,
+    "v": 36,
+    "w": 21,
+    "x": 41,
+    "y": 23,
+    "z": 31,
+    "|": 4,
+    "á": 33,
+    "é": 32,
+    "í": 30,
+    "ñ": 26,
+    "ó": 38,
+    "ø": 10,
+    "ú": 27,
+    "—": 29
+  },
+  "guo": {
+    "'": 28,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 19,
+    "d": 23,
+    "e": 6,
+    "f": 27,
+    "g": 33,
+    "h": 20,
+    "i": 8,
+    "j": 7,
+    "k": 21,
+    "l": 11,
+    "m": 17,
+    "n": 9,
+    "o": 12,
+    "p": 16,
+    "q": 39,
+    "r": 26,
+    "s": 14,
+    "t": 10,
+    "u": 25,
+    "v": 36,
+    "w": 15,
+    "x": 13,
+    "y": 29,
+    "z": 38,
+    "|": 5,
+    "á": 35,
+    "é": 34,
+    "í": 32,
+    "ñ": 40,
+    "ó": 37,
+    "ú": 30,
+    "ʉ": 18,
+    "—": 31
+  },
+  "guq": {
+    "'": 30,
+    "0": 43,
+    "1": 44,
+    "2": 45,
+    "5": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 22,
+    "d": 12,
+    "e": 6,
+    "f": 32,
+    "g": 17,
+    "h": 23,
+    "i": 11,
+    "j": 14,
+    "k": 21,
+    "l": 25,
+    "m": 19,
+    "n": 18,
+    "o": 9,
+    "p": 10,
+    "q": 35,
+    "r": 13,
+    "s": 27,
+    "t": 24,
+    "u": 7,
+    "w": 8,
+    "x": 34,
+    "y": 15,
+    "z": 36,
+    "|": 4,
+    "­": 47,
+    "á": 38,
+    "ã": 16,
+    "é": 46,
+    "í": 42,
+    "ñ": 37,
+    "ó": 40,
+    "õ": 26,
+    "ĩ": 29,
+    "ũ": 33,
+    "̃": 39,
+    "ẽ": 28,
+    "–": 41,
+    "—": 31
+  },
+  "guu": {
+    "'": 47,
+    "0": 31,
+    "1": 32,
+    "2": 34,
+    "3": 37,
+    "4": 35,
+    "5": 41,
+    "6": 43,
+    "7": 33,
+    "8": 44,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 8,
+    "c": 14,
+    "d": 18,
+    "e": 16,
+    "f": 22,
+    "g": 42,
+    "h": 25,
+    "i": 6,
+    "j": 15,
+    "l": 10,
+    "m": 12,
+    "n": 19,
+    "o": 9,
+    "p": 30,
+    "q": 26,
+    "r": 28,
+    "s": 24,
+    "t": 23,
+    "u": 13,
+    "v": 36,
+    "w": 17,
+    "y": 21,
+    "|": 4,
+    "á": 45,
+    "ä": 7,
+    "é": 38,
+    "í": 40,
+    "ñ": 48,
+    "ó": 39,
+    "ö": 11,
+    "ú": 29,
+    "̦": 20,
+    "—": 27
+  },
+  "gux": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 22,
+    "d": 13,
+    "e": 8,
+    "f": 23,
+    "g": 18,
+    "h": 27,
+    "i": 6,
+    "j": 21,
+    "k": 14,
+    "l": 9,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "s": 20,
+    "t": 17,
+    "u": 10,
+    "v": 29,
+    "w": 24,
+    "y": 15,
+    "|": 4,
+    "ñ": 26,
+    "ŋ": 25
+  },
+  "gvc": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 15,
+    "d": 22,
+    "e": 11,
+    "f": 35,
+    "g": 27,
+    "h": 10,
+    "i": 6,
+    "j": 18,
+    "k": 43,
+    "l": 31,
+    "m": 16,
+    "n": 14,
+    "o": 9,
+    "p": 20,
+    "q": 34,
+    "r": 7,
+    "s": 17,
+    "t": 13,
+    "u": 8,
+    "v": 37,
+    "w": 23,
+    "x": 45,
+    "y": 19,
+    "z": 39,
+    "|": 4,
+    "à": 44,
+    "á": 41,
+    "ã": 21,
+    "é": 38,
+    "í": 28,
+    "ñ": 25,
+    "ó": 33,
+    "õ": 29,
+    "ú": 42,
+    "ĩ": 32,
+    "ũ": 26,
+    "̶": 12,
+    "ẽ": 36,
+    "—": 30
+  },
+  "gvl": {
+    "'": 18,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 23,
+    "c": 37,
+    "d": 10,
+    "e": 5,
+    "f": 47,
+    "g": 12,
+    "h": 35,
+    "i": 9,
+    "j": 21,
+    "k": 8,
+    "l": 17,
+    "m": 16,
+    "n": 6,
+    "o": 13,
+    "p": 32,
+    "q": 44,
+    "r": 14,
+    "s": 20,
+    "t": 11,
+    "u": 15,
+    "v": 41,
+    "w": 28,
+    "x": 48,
+    "y": 29,
+    "z": 42,
+    "|": 4,
+    "à": 27,
+    "á": 30,
+    "è": 19,
+    "é": 22,
+    "ê": 31,
+    "í": 53,
+    "ò": 33,
+    "ó": 39,
+    "ù": 50,
+    "ú": 40,
+    "ĝ": 51,
+    "ĵ": 43,
+    "ń": 55,
+    "ŝ": 49,
+    "š": 52,
+    "ɓ": 25,
+    "ɔ": 24,
+    "ɗ": 38,
+    "̀": 46,
+    "́": 54,
+    "̂": 34,
+    "ḿ": 45,
+    "–": 36
+  },
+  "gwi": {
+    "'": 8,
+    "-": 35,
+    "0": 36,
+    "1": 37,
+    "2": 40,
+    "3": 43,
+    "4": 41,
+    "5": 44,
+    "6": 46,
+    "7": 42,
+    "8": 47,
+    "9": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 34,
+    "c": 22,
+    "d": 17,
+    "e": 11,
+    "f": 38,
+    "g": 12,
+    "h": 7,
+    "i": 5,
+    "j": 21,
+    "k": 16,
+    "l": 25,
+    "m": 32,
+    "n": 9,
+    "o": 20,
+    "p": 33,
+    "q": 45,
+    "r": 23,
+    "s": 18,
+    "t": 10,
+    "u": 27,
+    "v": 26,
+    "w": 14,
+    "x": 39,
+    "y": 19,
+    "z": 24,
+    "|": 4,
+    "ą": 13,
+    "ę": 29,
+    "į": 15,
+    "ł": 28,
+    "ų": 30,
+    "ǫ": 31
+  },
+  "gwr": {
+    "'": 40,
+    "-": 29,
+    "0": 30,
+    "1": 31,
+    "2": 34,
+    "3": 35,
+    "4": 32,
+    "5": 36,
+    "6": 37,
+    "7": 39,
+    "8": 41,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 9,
+    "c": 27,
+    "d": 23,
+    "e": 7,
+    "f": 25,
+    "g": 19,
+    "i": 6,
+    "j": 26,
+    "k": 12,
+    "l": 18,
+    "m": 13,
+    "n": 11,
+    "o": 8,
+    "p": 24,
+    "r": 16,
+    "s": 20,
+    "t": 17,
+    "u": 10,
+    "v": 28,
+    "w": 15,
+    "y": 14,
+    "z": 22,
+    "|": 5,
+    "ŋ": 33,
+    "ʼ": 21
+  },
+  "gym": {
+    "-": 47,
+    "0": 40,
+    "1": 38,
+    "2": 39,
+    "3": 46,
+    "4": 44,
+    "5": 45,
+    "6": 42,
+    "7": 50,
+    "8": 43,
+    "9": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 13,
+    "c": 28,
+    "d": 20,
+    "e": 5,
+    "f": 32,
+    "g": 19,
+    "h": 30,
+    "i": 9,
+    "j": 21,
+    "k": 8,
+    "l": 26,
+    "m": 14,
+    "n": 7,
+    "o": 17,
+    "p": 27,
+    "q": 37,
+    "r": 10,
+    "s": 23,
+    "t": 11,
+    "u": 16,
+    "v": 34,
+    "w": 15,
+    "x": 49,
+    "y": 22,
+    "z": 36,
+    "|": 4,
+    "á": 35,
+    "ä": 12,
+    "é": 29,
+    "í": 31,
+    "ñ": 25,
+    "ó": 33,
+    "ö": 18,
+    "ú": 41,
+    "ü": 24
+  },
+  "gyr": {
+    "'": 16,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 21,
+    "c": 26,
+    "d": 23,
+    "e": 5,
+    "f": 41,
+    "g": 25,
+    "h": 28,
+    "i": 9,
+    "j": 31,
+    "k": 15,
+    "l": 33,
+    "m": 18,
+    "n": 20,
+    "o": 7,
+    "p": 10,
+    "q": 45,
+    "r": 8,
+    "s": 11,
+    "t": 17,
+    "u": 13,
+    "v": 12,
+    "w": 22,
+    "x": 46,
+    "y": 19,
+    "z": 44,
+    "|": 4,
+    "á": 43,
+    "ä": 24,
+    "é": 40,
+    "ë": 29,
+    "í": 37,
+    "ï": 36,
+    "ñ": 27,
+    "ó": 42,
+    "ö": 34,
+    "ú": 35,
+    "ü": 30,
+    "ɨ": 14,
+    "̇": 32,
+    "–": 38
+  },
+  "had": {
+    "'": 29,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 14,
+    "c": 25,
+    "d": 15,
+    "e": 8,
+    "f": 27,
+    "g": 13,
+    "h": 19,
+    "i": 5,
+    "j": 23,
+    "k": 16,
+    "l": 22,
+    "m": 12,
+    "n": 6,
+    "o": 9,
+    "p": 20,
+    "r": 21,
+    "s": 18,
+    "t": 11,
+    "u": 10,
+    "w": 24,
+    "y": 17,
+    "z": 28,
+    "|": 4
+  },
+  "hag": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 31,
+    "a": 5,
+    "b": 12,
+    "c": 27,
+    "d": 20,
+    "e": 8,
+    "f": 28,
+    "g": 22,
+    "h": 25,
+    "i": 6,
+    "j": 26,
+    "k": 16,
+    "l": 17,
+    "m": 11,
+    "n": 7,
+    "o": 13,
+    "p": 21,
+    "q": 30,
+    "r": 15,
+    "s": 10,
+    "t": 18,
+    "u": 9,
+    "v": 29,
+    "w": 24,
+    "y": 14,
+    "z": 23,
+    "|": 4,
+    "ŋ": 19
+  },
+  "hak": {
+    "'": 45,
+    "-": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 41,
+    "a": 18,
+    "c": 14,
+    "e": 13,
+    "f": 33,
+    "g": 10,
+    "h": 7,
+    "i": 8,
+    "k": 9,
+    "l": 20,
+    "m": 21,
+    "n": 5,
+    "o": 16,
+    "p": 23,
+    "s": 15,
+    "t": 11,
+    "u": 12,
+    "v": 34,
+    "y": 19,
+    "|": 4,
+    "à": 29,
+    "á": 32,
+    "â": 22,
+    "è": 38,
+    "é": 39,
+    "ê": 24,
+    "ì": 17,
+    "í": 40,
+    "î": 28,
+    "ò": 30,
+    "ó": 25,
+    "ô": 35,
+    "ù": 37,
+    "ú": 36,
+    "û": 27,
+    "̀": 43,
+    "́": 42,
+    "̂": 44,
+    "̍": 31,
+    "ṳ": 26
+  },
+  "hap": {
+    "'": 24,
+    "-": 26,
+    "0": 27,
+    "1": 28,
+    "2": 31,
+    "3": 35,
+    "4": 32,
+    "5": 33,
+    "6": 36,
+    "7": 30,
+    "8": 38,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 39,
+    "d": 21,
+    "e": 6,
+    "f": 22,
+    "g": 10,
+    "h": 13,
+    "i": 7,
+    "j": 34,
+    "k": 11,
+    "l": 12,
+    "m": 15,
+    "n": 8,
+    "o": 9,
+    "p": 25,
+    "r": 19,
+    "s": 18,
+    "t": 16,
+    "u": 14,
+    "w": 17,
+    "y": 23,
+    "z": 29,
+    "|": 4
+  },
+  "hat": {
+    "'": 29,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 31,
+    "d": 20,
+    "e": 7,
+    "f": 24,
+    "g": 27,
+    "h": 30,
+    "i": 9,
+    "j": 25,
+    "k": 17,
+    "l": 11,
+    "m": 16,
+    "n": 5,
+    "o": 8,
+    "p": 12,
+    "r": 19,
+    "s": 15,
+    "t": 10,
+    "u": 13,
+    "v": 21,
+    "w": 23,
+    "y": 14,
+    "z": 28,
+    "|": 4,
+    "à": 33,
+    "è": 18,
+    "ò": 26,
+    "—": 32
+  },
+  "hau": {
+    "!": 74,
+    "$": 52,
+    "%": 56,
+    "&": 76,
+    "'": 28,
+    "+": 64,
+    ",": 45,
+    "-": 33,
+    ".": 43,
+    "/": 48,
+    "0": 30,
+    "1": 32,
+    "2": 34,
+    "3": 40,
+    "4": 39,
+    "5": 38,
+    "6": 42,
+    "7": 44,
+    "8": 41,
+    "9": 36,
+    ":": 50,
+    ";": 54,
+    "<": 65,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    ">": 66,
+    "[": 62,
+    "]": 63,
+    "a": 4,
+    "b": 19,
+    "c": 23,
+    "d": 12,
+    "e": 16,
+    "f": 24,
+    "g": 21,
+    "h": 20,
+    "i": 7,
+    "j": 26,
+    "k": 8,
+    "l": 22,
+    "m": 14,
+    "n": 6,
+    "o": 18,
+    "p": 31,
+    "q": 49,
+    "r": 10,
+    "s": 11,
+    "t": 15,
+    "u": 9,
+    "v": 35,
+    "w": 17,
+    "x": 46,
+    "y": 13,
+    "z": 25,
+    "|": 5,
+    "£": 71,
+    "¥": 68,
+    "°": 61,
+    "²": 72,
+    "á": 58,
+    "ã": 70,
+    "ç": 67,
+    "é": 69,
+    "í": 59,
+    "ó": 73,
+    "õ": 60,
+    "ö": 78,
+    "ú": 75,
+    "ü": 53,
+    "ā": 77,
+    "ă": 82,
+    "ū": 83,
+    "ƙ": 27,
+    "ƴ": 55,
+    "ɓ": 37,
+    "ɗ": 29,
+    "ʻ": 79,
+    "ˈ": 81,
+    "—": 80,
+    "‘": 57,
+    "’": 47,
+    "”": 51
+  },
+  "hay": {
+    "'": 22,
+    "0": 30,
+    "1": 31,
+    "2": 32,
+    "3": 34,
+    "4": 33,
+    "5": 35,
+    "6": 37,
+    "7": 38,
+    "8": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "c": 28,
+    "d": 24,
+    "e": 10,
+    "f": 26,
+    "g": 15,
+    "h": 19,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 20,
+    "m": 13,
+    "n": 8,
+    "o": 9,
+    "p": 27,
+    "r": 14,
+    "s": 21,
+    "t": 17,
+    "u": 7,
+    "v": 29,
+    "w": 16,
+    "y": 18,
+    "z": 23,
+    "|": 5
+  },
+  "heb": {
+    "!": 86,
+    "\"": 36,
+    "$": 78,
+    "%": 72,
+    "'": 34,
+    "+": 80,
+    ",": 56,
+    "-": 31,
+    ".": 42,
+    "/": 68,
+    "0": 30,
+    "1": 32,
+    "2": 35,
+    "3": 45,
+    "4": 41,
+    "5": 40,
+    "6": 44,
+    "7": 51,
+    "8": 46,
+    "9": 39,
+    ":": 63,
+    ";": 76,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 88,
+    "[": 82,
+    "]": 83,
+    "a": 38,
+    "b": 64,
+    "c": 57,
+    "d": 58,
+    "e": 49,
+    "f": 69,
+    "g": 59,
+    "h": 62,
+    "i": 52,
+    "j": 73,
+    "k": 67,
+    "l": 53,
+    "m": 60,
+    "n": 43,
+    "o": 47,
+    "p": 54,
+    "q": 85,
+    "r": 48,
+    "s": 55,
+    "t": 50,
+    "u": 61,
+    "v": 66,
+    "w": 71,
+    "x": 75,
+    "y": 70,
+    "z": 74,
+    "|": 4,
+    "¥": 84,
+    "í": 89,
+    "õ": 77,
+    "ּ": 79,
+    "־": 65,
+    "א": 13,
+    "ב": 12,
+    "ג": 24,
+    "ד": 18,
+    "ה": 7,
+    "ו": 6,
+    "ז": 28,
+    "ח": 20,
+    "ט": 25,
+    "י": 5,
+    "ך": 29,
+    "כ": 21,
+    "ל": 8,
+    "ם": 15,
+    "מ": 11,
+    "ן": 27,
+    "נ": 16,
+    "ס": 23,
+    "ע": 17,
+    "ף": 33,
+    "פ": 22,
+    "ץ": 37,
+    "צ": 26,
+    "ק": 19,
+    "ר": 9,
+    "ש": 14,
+    "ת": 10,
+    "–": 87,
+    "—": 81
+  },
+  "heh": {
+    "'": 26,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 25,
+    "d": 22,
+    "e": 8,
+    "f": 24,
+    "g": 15,
+    "h": 21,
+    "i": 6,
+    "j": 28,
+    "k": 12,
+    "l": 10,
+    "m": 13,
+    "n": 9,
+    "o": 16,
+    "p": 20,
+    "s": 18,
+    "t": 19,
+    "u": 7,
+    "v": 14,
+    "w": 11,
+    "y": 17,
+    "z": 27,
+    "|": 4
+  },
+  "hif": {
+    "'": 34,
+    "-": 33,
+    "0": 37,
+    "1": 39,
+    "2": 42,
+    "4": 38,
+    "6": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 24,
+    "d": 22,
+    "e": 6,
+    "f": 29,
+    "g": 19,
+    "h": 7,
+    "i": 8,
+    "j": 21,
+    "k": 9,
+    "l": 16,
+    "m": 13,
+    "n": 14,
+    "o": 15,
+    "p": 20,
+    "q": 35,
+    "r": 10,
+    "s": 12,
+    "t": 18,
+    "u": 11,
+    "v": 32,
+    "w": 25,
+    "x": 36,
+    "y": 23,
+    "z": 31,
+    "|": 4,
+    "ñ": 26,
+    "ḍ": 30,
+    "ṛ": 28,
+    "ṭ": 27,
+    "–": 41
+  },
+  "hig": {
+    "'": 35,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 31,
+    "d": 21,
+    "e": 24,
+    "f": 30,
+    "g": 18,
+    "h": 16,
+    "i": 8,
+    "j": 25,
+    "k": 12,
+    "l": 14,
+    "m": 7,
+    "n": 9,
+    "o": 37,
+    "p": 29,
+    "r": 17,
+    "s": 15,
+    "t": 10,
+    "u": 22,
+    "v": 26,
+    "w": 11,
+    "y": 13,
+    "z": 23,
+    "|": 4,
+    "á": 19,
+    "í": 38,
+    "ú": 36,
+    "ɓ": 33,
+    "ɗ": 32,
+    "ə": 5,
+    "́": 27,
+    "—": 34
+  },
+  "hil": {
+    "'": 42,
+    "-": 24,
+    "0": 27,
+    "1": 33,
+    "2": 34,
+    "3": 37,
+    "4": 35,
+    "5": 36,
+    "6": 40,
+    "7": 41,
+    "8": 43,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 26,
+    "d": 17,
+    "e": 23,
+    "f": 28,
+    "g": 7,
+    "h": 20,
+    "i": 8,
+    "j": 25,
+    "k": 11,
+    "l": 13,
+    "m": 12,
+    "n": 6,
+    "o": 10,
+    "p": 16,
+    "q": 31,
+    "r": 21,
+    "s": 9,
+    "t": 14,
+    "u": 15,
+    "v": 29,
+    "w": 22,
+    "x": 38,
+    "y": 18,
+    "z": 30,
+    "|": 5,
+    "ñ": 44,
+    "—": 32
+  },
+  "hin": {
+    "!": 111,
+    "$": 109,
+    "%": 101,
+    "'": 94,
+    "+": 106,
+    ",": 75,
+    "-": 54,
+    ".": 72,
+    "/": 86,
+    "0": 50,
+    "1": 53,
+    "2": 57,
+    "3": 67,
+    "4": 63,
+    "5": 62,
+    "6": 68,
+    "7": 74,
+    "8": 69,
+    "9": 61,
+    ":": 90,
+    ";": 112,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 124,
+    "a": 73,
+    "b": 98,
+    "c": 79,
+    "d": 97,
+    "e": 83,
+    "f": 92,
+    "g": 95,
+    "h": 100,
+    "i": 84,
+    "j": 108,
+    "k": 116,
+    "l": 81,
+    "m": 89,
+    "n": 82,
+    "o": 76,
+    "p": 87,
+    "q": 120,
+    "r": 85,
+    "s": 88,
+    "t": 80,
+    "u": 96,
+    "v": 107,
+    "w": 102,
+    "x": 105,
+    "y": 103,
+    "z": 117,
+    "|": 4,
+    "£": 121,
+    "¥": 104,
+    "°": 114,
+    "á": 113,
+    "õ": 118,
+    "ú": 115,
+    "ँ": 77,
+    "ं": 13,
+    "ः": 93,
+    "अ": 33,
+    "आ": 38,
+    "इ": 39,
+    "ई": 47,
+    "उ": 45,
+    "ऊ": 78,
+    "ऋ": 122,
+    "ए": 30,
+    "ऐ": 64,
+    "ऑ": 66,
+    "ओ": 55,
+    "औ": 41,
+    "क": 6,
+    "ख": 44,
+    "ग": 27,
+    "घ": 56,
+    "च": 36,
+    "छ": 49,
+    "ज": 23,
+    "झ": 65,
+    "ञ": 70,
+    "ट": 28,
+    "ठ": 58,
+    "ड": 34,
+    "ढ": 60,
+    "ण": 48,
+    "त": 16,
+    "थ": 37,
+    "द": 25,
+    "ध": 42,
+    "न": 12,
+    "ऩ": 126,
+    "प": 21,
+    "फ": 43,
+    "ब": 26,
+    "भ": 40,
+    "म": 17,
+    "य": 20,
+    "र": 8,
+    "ल": 19,
+    "व": 22,
+    "श": 32,
+    "ष": 46,
+    "स": 10,
+    "ह": 14,
+    "़": 31,
+    "ा": 5,
+    "ि": 11,
+    "ी": 15,
+    "ु": 29,
+    "ू": 35,
+    "ृ": 59,
+    "ॅ": 123,
+    "ॆ": 125,
+    "े": 7,
+    "ै": 24,
+    "ॉ": 51,
+    "ो": 18,
+    "ौ": 52,
+    "्": 9,
+    "।": 71,
+    "​": 91,
+    "‍": 110,
+    "—": 119,
+    "”": 99
+  },
+  "hlb": {
+    "'": 52,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 28,
+    "अ": 37,
+    "आ": 20,
+    "इ": 49,
+    "ई": 35,
+    "उ": 27,
+    "ऊ": 53,
+    "ए": 33,
+    "ओ": 45,
+    "क": 10,
+    "ख": 32,
+    "ग": 25,
+    "घ": 48,
+    "च": 22,
+    "छ": 44,
+    "ज": 21,
+    "झ": 50,
+    "ट": 29,
+    "ठ": 47,
+    "ड": 30,
+    "ढ": 51,
+    "त": 15,
+    "थ": 46,
+    "द": 26,
+    "ध": 42,
+    "न": 7,
+    "प": 24,
+    "फ": 43,
+    "ब": 17,
+    "भ": 31,
+    "म": 12,
+    "य": 19,
+    "र": 9,
+    "ल": 11,
+    "व": 34,
+    "स": 14,
+    "ह": 16,
+    "़": 41,
+    "ा": 5,
+    "ि": 23,
+    "ी": 18,
+    "ु": 13,
+    "ू": 39,
+    "े": 6,
+    "ो": 8,
+    "्": 38,
+    "‍": 40
+  },
+  "hlt": {
+    "'": 28,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 23,
+    "d": 21,
+    "e": 9,
+    "f": 29,
+    "g": 11,
+    "h": 6,
+    "i": 10,
+    "j": 26,
+    "k": 13,
+    "l": 12,
+    "m": 16,
+    "n": 7,
+    "o": 17,
+    "p": 18,
+    "r": 22,
+    "s": 19,
+    "t": 8,
+    "u": 14,
+    "v": 24,
+    "w": 15,
+    "y": 25,
+    "z": 30,
+    "|": 4
+  },
+  "hne": {
+    "'": 55,
+    "-": 49,
+    "0": 63,
+    "3": 68,
+    "4": 62,
+    "5": 64,
+    "6": 65,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 16,
+    "ः": 57,
+    "अ": 23,
+    "आ": 37,
+    "इ": 35,
+    "ई": 39,
+    "उ": 50,
+    "ऊ": 27,
+    "ए": 32,
+    "ओ": 28,
+    "क": 8,
+    "ख": 31,
+    "ग": 30,
+    "घ": 43,
+    "च": 34,
+    "छ": 45,
+    "ज": 24,
+    "झ": 48,
+    "ट": 44,
+    "ठ": 46,
+    "ड": 40,
+    "ढ": 52,
+    "ण": 59,
+    "त": 15,
+    "थ": 36,
+    "द": 25,
+    "ध": 41,
+    "न": 9,
+    "प": 18,
+    "फ": 51,
+    "ब": 17,
+    "भ": 38,
+    "म": 10,
+    "य": 19,
+    "र": 6,
+    "ल": 14,
+    "व": 21,
+    "स": 12,
+    "ह": 11,
+    "़": 42,
+    "ा": 5,
+    "ि": 13,
+    "ी": 20,
+    "ु": 26,
+    "ू": 29,
+    "ृ": 67,
+    "े": 7,
+    "ै": 54,
+    "ो": 22,
+    "ौ": 53,
+    "्": 33,
+    "०": 60,
+    "१": 66,
+    "४": 61,
+    "५": 70,
+    "६": 69,
+    "‍": 47,
+    "–": 56,
+    "—": 58
+  },
+  "hnn": {
+    "'": 23,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "d": 15,
+    "g": 9,
+    "h": 19,
+    "i": 7,
+    "k": 11,
+    "l": 17,
+    "m": 12,
+    "n": 6,
+    "p": 16,
+    "r": 20,
+    "s": 10,
+    "t": 13,
+    "u": 8,
+    "w": 21,
+    "y": 14,
+    "|": 5
+  },
+  "hns": {
+    "'": 26,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "d": 17,
+    "e": 6,
+    "f": 25,
+    "g": 20,
+    "h": 9,
+    "i": 7,
+    "j": 19,
+    "k": 8,
+    "l": 15,
+    "m": 18,
+    "n": 16,
+    "o": 10,
+    "p": 21,
+    "r": 11,
+    "s": 13,
+    "t": 12,
+    "u": 23,
+    "w": 22,
+    "y": 27,
+    "z": 31,
+    "|": 5,
+    "á": 30,
+    "é": 28,
+    "í": 32,
+    "ó": 29
+  },
+  "hoc": {
+    "'": 29,
+    "-": 35,
+    "4": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "i": 55,
+    "|": 4,
+    "ଁ": 43,
+    "ଂ": 44,
+    "ଃ": 11,
+    "ଅ": 31,
+    "ଆ": 23,
+    "ଇ": 19,
+    "ଉ": 39,
+    "ଏ": 27,
+    "କ": 10,
+    "ଖ": 46,
+    "ଗ": 20,
+    "ଙ": 30,
+    "ଚ": 34,
+    "ଜ": 26,
+    "ଞ": 36,
+    "ଟ": 37,
+    "ଡ": 22,
+    "ଢ": 56,
+    "ଣ": 33,
+    "ତ": 13,
+    "ଥ": 47,
+    "ଦ": 21,
+    "ନ": 8,
+    "ପ": 25,
+    "ଫ": 45,
+    "ବ": 24,
+    "ଭ": 54,
+    "ମ": 15,
+    "ର": 12,
+    "ଲ": 17,
+    "ଳ": 41,
+    "ଶ": 50,
+    "ସ": 16,
+    "ହ": 28,
+    "ା": 5,
+    "ି": 9,
+    "ୀ": 42,
+    "ୁ": 14,
+    "ୃ": 51,
+    "େ": 6,
+    "ୈ": 57,
+    "ୋ": 38,
+    "ୌ": 48,
+    "୍": 7,
+    "ୟ": 18,
+    "ୱ": 32,
+    "‍": 40,
+    "–": 53,
+    "—": 52
+  },
+  "hoy": {
+    "'": 63,
+    "-": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 62,
+    "ं": 22,
+    "अ": 23,
+    "आ": 19,
+    "इ": 36,
+    "ई": 31,
+    "उ": 37,
+    "ऊ": 33,
+    "ए": 59,
+    "ओ": 55,
+    "औ": 57,
+    "क": 10,
+    "ख": 43,
+    "ग": 24,
+    "घ": 46,
+    "च": 35,
+    "छ": 56,
+    "ज": 34,
+    "झ": 58,
+    "ट": 29,
+    "ठ": 53,
+    "ड": 41,
+    "ढ": 44,
+    "ण": 47,
+    "त": 13,
+    "थ": 51,
+    "द": 6,
+    "ध": 42,
+    "न": 8,
+    "प": 28,
+    "फ": 49,
+    "ब": 30,
+    "भ": 40,
+    "म": 18,
+    "य": 15,
+    "र": 7,
+    "ल": 25,
+    "ळ": 17,
+    "व": 26,
+    "श": 32,
+    "ष": 54,
+    "स": 14,
+    "ह": 27,
+    "़": 60,
+    "ा": 5,
+    "ि": 16,
+    "ी": 11,
+    "ु": 9,
+    "ू": 39,
+    "ृ": 61,
+    "े": 21,
+    "ै": 48,
+    "ॉ": 45,
+    "ो": 20,
+    "ौ": 50,
+    "्": 12,
+    "‍": 38
+  },
+  "hrv": {
+    "!": 60,
+    "\"": 34,
+    "%": 56,
+    "'": 52,
+    "+": 67,
+    ",": 50,
+    "-": 40,
+    ".": 32,
+    "/": 48,
+    "0": 31,
+    "1": 35,
+    "2": 36,
+    "3": 43,
+    "4": 42,
+    "5": 41,
+    "6": 44,
+    "7": 46,
+    "8": 45,
+    "9": 39,
+    ":": 47,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 72,
+    "[": 70,
+    "]": 71,
+    "a": 5,
+    "b": 23,
+    "c": 24,
+    "d": 18,
+    "e": 7,
+    "f": 30,
+    "g": 22,
+    "h": 26,
+    "i": 6,
+    "j": 10,
+    "k": 16,
+    "l": 15,
+    "m": 19,
+    "n": 9,
+    "o": 8,
+    "p": 20,
+    "q": 51,
+    "r": 11,
+    "s": 13,
+    "t": 12,
+    "u": 14,
+    "v": 17,
+    "w": 38,
+    "x": 49,
+    "y": 37,
+    "z": 21,
+    "|": 4,
+    "¥": 62,
+    "°": 59,
+    "²": 74,
+    "á": 61,
+    "ç": 63,
+    "é": 68,
+    "í": 58,
+    "ó": 75,
+    "õ": 64,
+    "ö": 65,
+    "ú": 69,
+    "ü": 57,
+    "ć": 29,
+    "č": 25,
+    "đ": 33,
+    "ł": 73,
+    "š": 27,
+    "ž": 28,
+    "​": 53,
+    "–": 55,
+    "”": 66
+  },
+  "hsb": {
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 23,
+    "d": 18,
+    "e": 6,
+    "f": 36,
+    "g": 35,
+    "h": 20,
+    "i": 12,
+    "j": 9,
+    "k": 15,
+    "l": 24,
+    "m": 17,
+    "n": 8,
+    "o": 7,
+    "p": 19,
+    "r": 14,
+    "s": 11,
+    "t": 13,
+    "u": 16,
+    "v": 40,
+    "w": 10,
+    "y": 21,
+    "z": 22,
+    "|": 4,
+    "á": 41,
+    "ó": 32,
+    "ü": 43,
+    "ć": 27,
+    "č": 33,
+    "ě": 26,
+    "ł": 29,
+    "ń": 37,
+    "ň": 42,
+    "ř": 34,
+    "š": 28,
+    "ź": 31,
+    "ž": 30,
+    "–": 39
+  },
+  "hto": {
+    "'": 37,
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 39,
+    "a": 5,
+    "b": 25,
+    "c": 14,
+    "d": 13,
+    "e": 7,
+    "f": 17,
+    "g": 28,
+    "h": 32,
+    "i": 8,
+    "j": 18,
+    "k": 35,
+    "l": 15,
+    "m": 11,
+    "n": 10,
+    "o": 9,
+    "p": 31,
+    "q": 36,
+    "r": 16,
+    "s": 29,
+    "t": 20,
+    "u": 12,
+    "v": 33,
+    "x": 40,
+    "y": 38,
+    "z": 22,
+    "|": 4,
+    "à": 27,
+    "á": 21,
+    "é": 26,
+    "í": 19,
+    "ñ": 24,
+    "ó": 23,
+    "ù": 6,
+    "ú": 30,
+    "ü": 34
+  },
+  "hub": {
+    "'": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 16,
+    "d": 25,
+    "e": 14,
+    "f": 31,
+    "g": 21,
+    "h": 15,
+    "i": 6,
+    "j": 17,
+    "k": 10,
+    "l": 23,
+    "m": 11,
+    "n": 8,
+    "o": 22,
+    "p": 19,
+    "q": 35,
+    "r": 13,
+    "s": 12,
+    "t": 9,
+    "u": 7,
+    "v": 30,
+    "w": 18,
+    "x": 36,
+    "y": 20,
+    "z": 33,
+    "|": 5,
+    "á": 28,
+    "é": 29,
+    "í": 26,
+    "ñ": 37,
+    "ó": 32,
+    "ú": 24
+  },
+  "hui": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "d": 16,
+    "e": 9,
+    "g": 10,
+    "h": 14,
+    "i": 6,
+    "k": 21,
+    "l": 12,
+    "m": 15,
+    "n": 7,
+    "o": 8,
+    "p": 22,
+    "r": 17,
+    "s": 23,
+    "t": 19,
+    "u": 13,
+    "w": 18,
+    "y": 20,
+    "|": 5,
+    "á": 27,
+    "é": 29,
+    "í": 24,
+    "̠": 26,
+    "̱": 25
+  },
+  "hun": {
+    "!": 63,
+    "\"": 42,
+    "%": 58,
+    "'": 57,
+    "+": 65,
+    ",": 53,
+    "-": 39,
+    ".": 41,
+    "/": 52,
+    "0": 36,
+    "1": 38,
+    "2": 40,
+    "3": 47,
+    "4": 46,
+    "5": 45,
+    "6": 48,
+    "7": 51,
+    "8": 49,
+    "9": 44,
+    ":": 56,
+    ";": 64,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 73,
+    "[": 69,
+    "]": 70,
+    "a": 6,
+    "b": 21,
+    "c": 32,
+    "d": 23,
+    "e": 5,
+    "f": 30,
+    "g": 17,
+    "h": 24,
+    "i": 15,
+    "j": 26,
+    "k": 11,
+    "l": 8,
+    "m": 19,
+    "n": 10,
+    "o": 12,
+    "p": 29,
+    "q": 54,
+    "r": 13,
+    "s": 9,
+    "t": 7,
+    "u": 28,
+    "v": 22,
+    "w": 43,
+    "x": 50,
+    "y": 20,
+    "z": 14,
+    "|": 4,
+    "°": 59,
+    "²": 71,
+    "×": 74,
+    "á": 16,
+    "ã": 67,
+    "ç": 66,
+    "é": 18,
+    "í": 33,
+    "ó": 27,
+    "õ": 60,
+    "ö": 25,
+    "ú": 35,
+    "û": 75,
+    "ü": 34,
+    "ő": 31,
+    "ű": 37,
+    "̇": 72,
+    "​": 68,
+    "–": 61,
+    "—": 62,
+    "”": 55
+  },
+  "hus-dialect_centralveracruz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 19,
+    "d": 23,
+    "e": 12,
+    "f": 28,
+    "g": 27,
+    "h": 15,
+    "i": 6,
+    "j": 11,
+    "k": 9,
+    "l": 10,
+    "m": 21,
+    "n": 8,
+    "o": 16,
+    "p": 24,
+    "r": 25,
+    "s": 13,
+    "t": 7,
+    "u": 17,
+    "w": 20,
+    "x": 18,
+    "y": 22,
+    "|": 4,
+    "ß": 31,
+    "á": 29,
+    "ó": 30,
+    "œ": 32,
+    "—": 26
+  },
+  "hus-dialect_westernpotosino": {
+    "'": 10,
+    "-": 39,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "3": 43,
+    "4": 40,
+    "5": 42,
+    "6": 41,
+    "7": 47,
+    "9": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 21,
+    "d": 25,
+    "e": 16,
+    "f": 33,
+    "g": 32,
+    "h": 14,
+    "i": 6,
+    "j": 12,
+    "k": 11,
+    "l": 9,
+    "m": 20,
+    "n": 8,
+    "o": 18,
+    "p": 26,
+    "r": 31,
+    "s": 17,
+    "t": 7,
+    "u": 19,
+    "v": 34,
+    "w": 23,
+    "x": 22,
+    "y": 24,
+    "z": 35,
+    "|": 4,
+    "à": 46,
+    "á": 13,
+    "é": 27,
+    "í": 28,
+    "ñ": 45,
+    "ó": 29,
+    "ú": 30
+  },
+  "huu": {
+    "'": 39,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 12,
+    "d": 15,
+    "e": 8,
+    "f": 21,
+    "g": 23,
+    "h": 32,
+    "i": 6,
+    "j": 17,
+    "k": 38,
+    "l": 14,
+    "m": 11,
+    "n": 10,
+    "o": 9,
+    "p": 26,
+    "q": 35,
+    "r": 16,
+    "s": 19,
+    "t": 18,
+    "u": 13,
+    "v": 30,
+    "x": 37,
+    "y": 29,
+    "z": 33,
+    "|": 5,
+    "á": 31,
+    "é": 27,
+    "í": 28,
+    "ñ": 20,
+    "ó": 34,
+    "ú": 25,
+    "ɨ": 7,
+    "—": 24
+  },
+  "huv": {
+    "1": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 15,
+    "d": 21,
+    "e": 8,
+    "f": 33,
+    "g": 18,
+    "h": 27,
+    "i": 7,
+    "j": 12,
+    "k": 38,
+    "l": 16,
+    "m": 11,
+    "n": 6,
+    "o": 13,
+    "p": 20,
+    "q": 25,
+    "r": 23,
+    "s": 17,
+    "t": 9,
+    "u": 22,
+    "v": 35,
+    "w": 14,
+    "x": 24,
+    "y": 19,
+    "z": 36,
+    "|": 4,
+    "á": 28,
+    "é": 32,
+    "í": 29,
+    "ñ": 37,
+    "ó": 31,
+    "ú": 30,
+    "ü": 10,
+    "ǘ": 34
+  },
+  "hvn": {
+    "'": 13,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "d": 11,
+    "e": 6,
+    "f": 28,
+    "g": 21,
+    "h": 20,
+    "i": 7,
+    "j": 22,
+    "k": 18,
+    "l": 17,
+    "m": 19,
+    "n": 10,
+    "o": 8,
+    "p": 16,
+    "r": 12,
+    "s": 23,
+    "t": 15,
+    "u": 9,
+    "w": 24,
+    "y": 26,
+    "|": 4,
+    "è": 14,
+    "ó": 29
+  },
+  "hwc": {
+    "'": 36,
+    "-": 33,
+    "0": 32,
+    "1": 34,
+    "2": 35,
+    "4": 37,
+    "5": 39,
+    "6": 38,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "c": 24,
+    "d": 8,
+    "e": 5,
+    "f": 20,
+    "g": 15,
+    "h": 17,
+    "i": 12,
+    "j": 27,
+    "k": 22,
+    "l": 16,
+    "m": 18,
+    "n": 9,
+    "o": 7,
+    "p": 23,
+    "q": 31,
+    "r": 19,
+    "s": 11,
+    "t": 10,
+    "u": 13,
+    "v": 26,
+    "w": 21,
+    "x": 29,
+    "y": 14,
+    "z": 28,
+    "|": 4,
+    "—": 30
+  },
+  "hye": {
+    "!": 96,
+    "%": 82,
+    "'": 77,
+    "+": 92,
+    ",": 59,
+    "-": 44,
+    ".": 61,
+    "/": 65,
+    "0": 43,
+    "1": 45,
+    "2": 47,
+    "3": 52,
+    "4": 51,
+    "5": 50,
+    "6": 54,
+    "7": 57,
+    "8": 53,
+    "9": 49,
+    ":": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 99,
+    "]": 100,
+    "`": 72,
+    "a": 63,
+    "b": 84,
+    "c": 70,
+    "d": 71,
+    "e": 67,
+    "f": 83,
+    "g": 74,
+    "h": 76,
+    "i": 66,
+    "j": 91,
+    "k": 88,
+    "l": 78,
+    "m": 86,
+    "n": 62,
+    "o": 64,
+    "p": 68,
+    "r": 69,
+    "s": 75,
+    "t": 73,
+    "u": 79,
+    "v": 80,
+    "w": 87,
+    "x": 89,
+    "y": 94,
+    "|": 4,
+    "«": 60,
+    "°": 97,
+    "»": 58,
+    "í": 101,
+    "ó": 102,
+    "õ": 93,
+    "՛": 85,
+    "՝": 55,
+    "ա": 5,
+    "բ": 28,
+    "գ": 25,
+    "դ": 24,
+    "ե": 9,
+    "զ": 34,
+    "է": 29,
+    "ը": 23,
+    "թ": 21,
+    "ժ": 40,
+    "ի": 10,
+    "լ": 18,
+    "խ": 35,
+    "ծ": 30,
+    "կ": 13,
+    "հ": 19,
+    "ձ": 39,
+    "ղ": 26,
+    "ճ": 42,
+    "մ": 12,
+    "յ": 15,
+    "ն": 6,
+    "շ": 31,
+    "ո": 8,
+    "չ": 36,
+    "պ": 27,
+    "ջ": 37,
+    "ռ": 32,
+    "ս": 17,
+    "վ": 16,
+    "տ": 14,
+    "ր": 7,
+    "ց": 20,
+    "ւ": 11,
+    "փ": 38,
+    "ք": 22,
+    "օ": 41,
+    "ֆ": 46,
+    "և": 33,
+    "։": 56,
+    "​": 95,
+    "—": 98,
+    "’": 90,
+    "․": 81
+  },
+  "hyw": {
+    "'": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ա": 5,
+    "բ": 28,
+    "գ": 31,
+    "դ": 27,
+    "ե": 9,
+    "զ": 26,
+    "է": 13,
+    "ը": 16,
+    "թ": 25,
+    "ժ": 41,
+    "ի": 11,
+    "լ": 20,
+    "խ": 34,
+    "ծ": 30,
+    "կ": 17,
+    "հ": 22,
+    "ձ": 36,
+    "ղ": 29,
+    "ճ": 42,
+    "մ": 14,
+    "յ": 19,
+    "ն": 6,
+    "շ": 35,
+    "ո": 7,
+    "չ": 32,
+    "պ": 24,
+    "ջ": 37,
+    "ռ": 33,
+    "ս": 12,
+    "վ": 23,
+    "տ": 15,
+    "ր": 8,
+    "ց": 21,
+    "ւ": 10,
+    "փ": 40,
+    "ք": 18,
+    "օ": 38,
+    "ֆ": 43
+  },
+  "iba": {
+    "-": 26,
+    "0": 28,
+    "2": 31,
+    "3": 29,
+    "5": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 25,
+    "d": 14,
+    "e": 9,
+    "g": 11,
+    "h": 19,
+    "i": 6,
+    "j": 22,
+    "k": 10,
+    "l": 13,
+    "m": 16,
+    "n": 7,
+    "o": 23,
+    "p": 21,
+    "r": 17,
+    "s": 15,
+    "t": 12,
+    "u": 8,
+    "v": 27,
+    "w": 24,
+    "y": 20,
+    "|": 5
+  },
+  "ibo": {
+    "!": 83,
+    "\"": 78,
+    "$": 52,
+    "%": 70,
+    "&": 87,
+    "'": 27,
+    "+": 63,
+    ",": 47,
+    "-": 29,
+    ".": 43,
+    "/": 50,
+    "0": 33,
+    "1": 34,
+    "2": 36,
+    "3": 41,
+    "4": 39,
+    "5": 38,
+    "6": 40,
+    "7": 45,
+    "8": 42,
+    "9": 37,
+    ":": 48,
+    ";": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 74,
+    "]": 75,
+    "_": 80,
+    "a": 5,
+    "b": 18,
+    "c": 28,
+    "d": 22,
+    "e": 6,
+    "f": 32,
+    "g": 17,
+    "h": 16,
+    "i": 8,
+    "j": 31,
+    "k": 9,
+    "l": 24,
+    "m": 12,
+    "n": 7,
+    "o": 13,
+    "p": 26,
+    "q": 49,
+    "r": 11,
+    "s": 23,
+    "t": 20,
+    "u": 15,
+    "v": 35,
+    "w": 21,
+    "x": 46,
+    "y": 25,
+    "z": 30,
+    "|": 4,
+    "£": 82,
+    "¥": 69,
+    "°": 68,
+    "²": 88,
+    "à": 57,
+    "á": 67,
+    "ã": 72,
+    "ç": 61,
+    "è": 56,
+    "é": 65,
+    "ì": 71,
+    "í": 73,
+    "ñ": 51,
+    "ò": 54,
+    "ó": 79,
+    "õ": 64,
+    "ù": 66,
+    "ú": 81,
+    "ü": 58,
+    "ē": 85,
+    "ł": 86,
+    "ū": 77,
+    "̀": 84,
+    "̄": 53,
+    "ṅ": 60,
+    "ị": 19,
+    "ọ": 14,
+    "ụ": 10,
+    "–": 59,
+    "—": 76,
+    "’": 44,
+    "”": 62
+  },
+  "icr": {
+    "'": 29,
+    "-": 31,
+    "0": 34,
+    "1": 35,
+    "2": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 28,
+    "d": 10,
+    "e": 8,
+    "f": 22,
+    "g": 19,
+    "h": 9,
+    "i": 6,
+    "j": 26,
+    "k": 21,
+    "l": 17,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 20,
+    "q": 33,
+    "r": 18,
+    "s": 13,
+    "t": 14,
+    "u": 12,
+    "v": 25,
+    "w": 15,
+    "x": 32,
+    "y": 24,
+    "z": 27,
+    "|": 4,
+    "—": 30
+  },
+  "idd": {
+    "'": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 12,
+    "b": 20,
+    "d": 38,
+    "e": 33,
+    "f": 36,
+    "g": 32,
+    "h": 23,
+    "i": 18,
+    "j": 30,
+    "k": 11,
+    "l": 14,
+    "m": 29,
+    "n": 5,
+    "o": 34,
+    "p": 39,
+    "r": 16,
+    "s": 13,
+    "t": 31,
+    "u": 27,
+    "w": 21,
+    "y": 22,
+    "|": 4,
+    "à": 10,
+    "á": 17,
+    "è": 26,
+    "é": 28,
+    "ì": 19,
+    "í": 6,
+    "ò": 35,
+    "ó": 24,
+    "ù": 37,
+    "ú": 25,
+    "ǹ": 40,
+    "ɔ": 7,
+    "ɛ": 8,
+    "̀": 15,
+    "́": 9,
+    "ḿ": 42
+  },
+  "ifa": {
+    "'": 16,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 9,
+    "e": 19,
+    "f": 28,
+    "g": 15,
+    "h": 11,
+    "i": 7,
+    "j": 24,
+    "k": 23,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 12,
+    "p": 18,
+    "r": 25,
+    "s": 21,
+    "t": 10,
+    "u": 8,
+    "v": 27,
+    "w": 22,
+    "x": 29,
+    "y": 14,
+    "z": 30,
+    "|": 5
+  },
+  "ifb": {
+    "'": 18,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 26,
+    "d": 9,
+    "e": 20,
+    "f": 29,
+    "g": 15,
+    "h": 11,
+    "i": 7,
+    "j": 25,
+    "k": 23,
+    "l": 16,
+    "m": 13,
+    "n": 6,
+    "o": 14,
+    "p": 17,
+    "q": 32,
+    "r": 24,
+    "s": 21,
+    "t": 12,
+    "u": 8,
+    "v": 27,
+    "w": 22,
+    "x": 30,
+    "y": 10,
+    "z": 28,
+    "|": 4
+  },
+  "ife": {
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 17,
+    "d": 35,
+    "e": 37,
+    "f": 32,
+    "g": 30,
+    "h": 50,
+    "i": 24,
+    "k": 10,
+    "l": 36,
+    "m": 31,
+    "n": 11,
+    "o": 38,
+    "p": 39,
+    "r": 25,
+    "s": 15,
+    "t": 13,
+    "u": 42,
+    "w": 22,
+    "y": 19,
+    "z": 34,
+    "|": 4,
+    "à": 9,
+    "á": 16,
+    "ã": 40,
+    "è": 26,
+    "é": 23,
+    "ì": 33,
+    "í": 6,
+    "ò": 29,
+    "ó": 20,
+    "ù": 44,
+    "ú": 28,
+    "ĩ": 48,
+    "ń": 41,
+    "ŋ": 18,
+    "ũ": 47,
+    "ǹ": 45,
+    "ɔ": 12,
+    "ɖ": 27,
+    "ɛ": 5,
+    "̀": 8,
+    "́": 7,
+    "̃": 43,
+    "ḿ": 51,
+    "ṹ": 49,
+    "‐": 46
+  },
+  "ifk": {
+    "'": 28,
+    "-": 22,
+    "1": 34,
+    "7": 36,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 27,
+    "a": 5,
+    "b": 20,
+    "c": 26,
+    "d": 9,
+    "e": 18,
+    "f": 29,
+    "g": 16,
+    "h": 12,
+    "i": 7,
+    "j": 25,
+    "k": 15,
+    "l": 17,
+    "m": 14,
+    "n": 6,
+    "o": 13,
+    "p": 19,
+    "q": 33,
+    "r": 24,
+    "s": 21,
+    "t": 10,
+    "u": 8,
+    "v": 30,
+    "w": 23,
+    "x": 32,
+    "y": 11,
+    "z": 31,
+    "|": 4
+  },
+  "ifu": {
+    "'": 26,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "c": 14,
+    "d": 23,
+    "e": 20,
+    "f": 25,
+    "g": 16,
+    "h": 7,
+    "i": 8,
+    "j": 15,
+    "k": 19,
+    "l": 18,
+    "m": 12,
+    "n": 6,
+    "o": 10,
+    "p": 13,
+    "r": 24,
+    "s": 22,
+    "t": 11,
+    "u": 9,
+    "v": 29,
+    "w": 27,
+    "x": 31,
+    "y": 17,
+    "z": 30,
+    "|": 4,
+    "á": 32
+  },
+  "ify": {
+    "'": 27,
+    "-": 19,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 26,
+    "d": 10,
+    "e": 7,
+    "f": 29,
+    "g": 12,
+    "h": 14,
+    "i": 8,
+    "j": 24,
+    "k": 17,
+    "l": 16,
+    "m": 13,
+    "n": 6,
+    "o": 23,
+    "p": 18,
+    "q": 32,
+    "r": 25,
+    "s": 20,
+    "t": 11,
+    "u": 9,
+    "v": 28,
+    "w": 22,
+    "x": 31,
+    "y": 15,
+    "z": 30,
+    "|": 4
+  },
+  "ign": {
+    "'": 11,
+    "-": 38,
+    "0": 40,
+    "1": 41,
+    "2": 44,
+    "3": 45,
+    "4": 42,
+    "9": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 17,
+    "d": 30,
+    "e": 7,
+    "f": 34,
+    "g": 33,
+    "h": 19,
+    "i": 6,
+    "j": 22,
+    "k": 13,
+    "l": 28,
+    "m": 14,
+    "n": 9,
+    "o": 26,
+    "p": 15,
+    "q": 37,
+    "r": 10,
+    "s": 16,
+    "t": 8,
+    "u": 12,
+    "v": 23,
+    "w": 18,
+    "x": 39,
+    "y": 21,
+    "z": 36,
+    "|": 5,
+    "á": 20,
+    "é": 24,
+    "í": 25,
+    "ñ": 29,
+    "ó": 35,
+    "ú": 27,
+    "–": 32
+  },
+  "ikk": {
+    "'": 28,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 24,
+    "c": 33,
+    "d": 23,
+    "e": 12,
+    "f": 32,
+    "g": 20,
+    "h": 14,
+    "i": 7,
+    "j": 30,
+    "k": 13,
+    "l": 21,
+    "m": 15,
+    "n": 5,
+    "o": 16,
+    "p": 29,
+    "r": 19,
+    "s": 22,
+    "t": 31,
+    "u": 18,
+    "v": 34,
+    "w": 17,
+    "y": 25,
+    "z": 27,
+    "|": 4,
+    "ò": 36,
+    "ù": 37,
+    "̀": 38,
+    "ṅ": 35,
+    "ẹ": 6,
+    "ị": 11,
+    "ọ": 10,
+    "ụ": 9
+  },
+  "ilb": {
+    "'": 29,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 10,
+    "c": 20,
+    "d": 22,
+    "e": 13,
+    "f": 26,
+    "g": 24,
+    "h": 18,
+    "i": 7,
+    "j": 25,
+    "k": 8,
+    "l": 12,
+    "m": 11,
+    "n": 9,
+    "o": 14,
+    "p": 23,
+    "s": 17,
+    "t": 16,
+    "u": 6,
+    "v": 28,
+    "w": 15,
+    "y": 21,
+    "z": 19,
+    "|": 5,
+    "ŋ": 27
+  },
+  "ilo": {
+    "'": 32,
+    "-": 24,
+    "0": 31,
+    "1": 34,
+    "2": 33,
+    "3": 37,
+    "4": 35,
+    "5": 38,
+    "6": 39,
+    "7": 41,
+    "8": 40,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 25,
+    "d": 11,
+    "e": 15,
+    "f": 27,
+    "g": 9,
+    "h": 26,
+    "i": 6,
+    "j": 23,
+    "k": 10,
+    "l": 19,
+    "m": 14,
+    "n": 7,
+    "o": 12,
+    "p": 17,
+    "q": 43,
+    "r": 20,
+    "s": 13,
+    "t": 8,
+    "u": 16,
+    "v": 29,
+    "w": 22,
+    "x": 36,
+    "y": 18,
+    "z": 30,
+    "|": 5,
+    "—": 28
+  },
+  "imo": {
+    "'": 32,
+    "0": 24,
+    "1": 28,
+    "2": 29,
+    "4": 31,
+    "5": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 17,
+    "d": 20,
+    "e": 9,
+    "f": 25,
+    "g": 15,
+    "h": 27,
+    "i": 5,
+    "j": 23,
+    "k": 13,
+    "l": 12,
+    "m": 11,
+    "n": 6,
+    "o": 7,
+    "p": 14,
+    "r": 19,
+    "s": 22,
+    "t": 16,
+    "u": 10,
+    "v": 26,
+    "w": 21,
+    "y": 18,
+    "|": 4
+  },
+  "ina": {
+    "'": 32,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 15,
+    "d": 18,
+    "e": 5,
+    "f": 22,
+    "g": 23,
+    "h": 21,
+    "i": 7,
+    "j": 25,
+    "k": 28,
+    "l": 13,
+    "m": 17,
+    "n": 10,
+    "o": 8,
+    "p": 16,
+    "q": 24,
+    "r": 12,
+    "s": 11,
+    "t": 9,
+    "u": 14,
+    "v": 19,
+    "w": 30,
+    "x": 26,
+    "y": 27,
+    "z": 31,
+    "|": 4,
+    "é": 33,
+    "ö": 34
+  },
+  "inb": {
+    "'": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 17,
+    "d": 21,
+    "e": 24,
+    "f": 32,
+    "g": 19,
+    "h": 18,
+    "i": 6,
+    "j": 23,
+    "k": 9,
+    "l": 16,
+    "m": 13,
+    "n": 10,
+    "o": 26,
+    "p": 12,
+    "r": 14,
+    "s": 11,
+    "t": 15,
+    "u": 8,
+    "v": 34,
+    "w": 20,
+    "x": 38,
+    "z": 33,
+    "|": 5,
+    "­": 7,
+    "á": 29,
+    "é": 31,
+    "í": 30,
+    "ñ": 22,
+    "ó": 35,
+    "ú": 28,
+    "̈": 36,
+    "—": 27
+  },
+  "ind": {
+    "!": 70,
+    "\"": 46,
+    "$": 57,
+    "%": 52,
+    "&": 64,
+    "'": 44,
+    "+": 66,
+    ",": 43,
+    "-": 27,
+    ".": 33,
+    "/": 42,
+    "0": 29,
+    "1": 30,
+    "2": 31,
+    "3": 36,
+    "4": 37,
+    "5": 35,
+    "6": 39,
+    "7": 40,
+    "8": 38,
+    "9": 34,
+    ":": 50,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "\\": 72,
+    "a": 4,
+    "b": 19,
+    "c": 24,
+    "d": 15,
+    "e": 7,
+    "f": 26,
+    "g": 16,
+    "h": 20,
+    "i": 8,
+    "j": 23,
+    "k": 12,
+    "l": 17,
+    "m": 14,
+    "n": 6,
+    "o": 21,
+    "p": 18,
+    "q": 47,
+    "r": 10,
+    "s": 13,
+    "t": 9,
+    "u": 11,
+    "v": 28,
+    "w": 25,
+    "x": 41,
+    "y": 22,
+    "z": 32,
+    "|": 5,
+    "¡": 51,
+    "£": 58,
+    "¥": 54,
+    "§": 59,
+    "©": 62,
+    "­": 56,
+    "°": 55,
+    "²": 63,
+    "µ": 68,
+    "¶": 69,
+    "º": 60,
+    "á": 73,
+    "â": 48,
+    "ã": 45,
+    "ä": 65,
+    "é": 75,
+    "ł": 77,
+    "ń": 78,
+    "ō": 76,
+    "ʼ": 79,
+    "–": 74,
+    "—": 71,
+    "”": 61,
+    "•": 67,
+    "€": 49
+  },
+  "iou": {
+    "'": 40,
+    "-": 25,
+    "0": 30,
+    "1": 29,
+    "2": 31,
+    "3": 36,
+    "4": 33,
+    "5": 35,
+    "6": 37,
+    "7": 34,
+    "8": 38,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "d": 21,
+    "e": 13,
+    "f": 28,
+    "g": 22,
+    "h": 26,
+    "i": 10,
+    "j": 24,
+    "k": 7,
+    "l": 27,
+    "m": 14,
+    "n": 8,
+    "o": 16,
+    "p": 15,
+    "r": 17,
+    "s": 23,
+    "t": 9,
+    "u": 12,
+    "v": 32,
+    "w": 18,
+    "y": 19,
+    "|": 4,
+    "ä": 5,
+    "ŋ": 11
+  },
+  "ipi": {
+    "-": 20,
+    "0": 25,
+    "2": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "d": 17,
+    "e": 6,
+    "g": 22,
+    "i": 9,
+    "j": 23,
+    "k": 13,
+    "l": 11,
+    "m": 14,
+    "n": 7,
+    "o": 8,
+    "p": 12,
+    "s": 21,
+    "t": 10,
+    "u": 15,
+    "w": 18,
+    "y": 16,
+    "|": 5,
+    "ŋ": 24
+  },
+  "iqw": {
+    "'": 25,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 13,
+    "c": 30,
+    "d": 23,
+    "e": 5,
+    "f": 31,
+    "g": 27,
+    "h": 11,
+    "i": 9,
+    "j": 29,
+    "k": 12,
+    "l": 17,
+    "m": 21,
+    "n": 8,
+    "o": 10,
+    "p": 18,
+    "r": 19,
+    "s": 26,
+    "t": 22,
+    "u": 6,
+    "v": 35,
+    "w": 24,
+    "y": 16,
+    "z": 32,
+    "|": 4,
+    "à": 42,
+    "á": 34,
+    "è": 44,
+    "é": 37,
+    "ì": 45,
+    "í": 36,
+    "ò": 46,
+    "ó": 38,
+    "ù": 43,
+    "ú": 39,
+    "ǹ": 47,
+    "̀": 41,
+    "́": 40,
+    "ẹ": 15,
+    "ị": 33,
+    "ọ": 20,
+    "ụ": 14
+  },
+  "iri": {
+    "'": 34,
+    "-": 29,
+    "3": 45,
+    "6": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 16,
+    "c": 22,
+    "d": 24,
+    "e": 5,
+    "f": 33,
+    "g": 13,
+    "h": 12,
+    "i": 7,
+    "j": 26,
+    "k": 17,
+    "l": 28,
+    "m": 20,
+    "n": 6,
+    "o": 19,
+    "p": 31,
+    "r": 9,
+    "s": 18,
+    "t": 23,
+    "u": 15,
+    "v": 32,
+    "w": 11,
+    "y": 14,
+    "z": 21,
+    "|": 4,
+    "á": 27,
+    "â": 40,
+    "è": 38,
+    "é": 25,
+    "ê": 37,
+    "í": 30,
+    "î": 36,
+    "ó": 35,
+    "ú": 39,
+    "ā": 42,
+    "ē": 41,
+    "ī": 43,
+    "̱": 10
+  },
+  "irk": {
+    "'": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 33,
+    "d": 22,
+    "e": 7,
+    "f": 31,
+    "g": 18,
+    "h": 15,
+    "i": 6,
+    "j": 34,
+    "k": 19,
+    "l": 16,
+    "m": 14,
+    "n": 9,
+    "o": 11,
+    "p": 32,
+    "q": 29,
+    "r": 8,
+    "s": 12,
+    "t": 17,
+    "u": 10,
+    "w": 20,
+    "x": 30,
+    "y": 24,
+    "z": 35,
+    "|": 4,
+    "á": 13,
+    "é": 28,
+    "í": 23,
+    "ó": 25,
+    "ú": 26
+  },
+  "isl": {
+    "%": 57,
+    "'": 56,
+    ",": 55,
+    "-": 39,
+    ".": 40,
+    "/": 54,
+    "0": 36,
+    "1": 37,
+    "2": 41,
+    "3": 46,
+    "4": 45,
+    "5": 44,
+    "6": 48,
+    "7": 50,
+    "8": 47,
+    "9": 42,
+    ":": 53,
+    ";": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 62,
+    "]": 63,
+    "a": 5,
+    "b": 27,
+    "c": 38,
+    "d": 22,
+    "e": 9,
+    "f": 17,
+    "g": 15,
+    "h": 21,
+    "i": 8,
+    "j": 26,
+    "k": 18,
+    "l": 13,
+    "m": 16,
+    "n": 7,
+    "o": 20,
+    "p": 30,
+    "q": 52,
+    "r": 6,
+    "s": 10,
+    "t": 11,
+    "u": 12,
+    "v": 19,
+    "w": 43,
+    "x": 49,
+    "y": 28,
+    "z": 51,
+    "|": 4,
+    "²": 60,
+    "á": 24,
+    "ã": 58,
+    "æ": 32,
+    "é": 35,
+    "í": 23,
+    "ð": 14,
+    "ó": 29,
+    "ö": 31,
+    "ú": 33,
+    "ý": 34,
+    "þ": 25,
+    "–": 61
+  },
+  "ita": {
+    "\"": 59,
+    "%": 52,
+    "&": 61,
+    "'": 25,
+    "+": 62,
+    ",": 53,
+    "-": 45,
+    ".": 41,
+    "/": 50,
+    "0": 29,
+    "1": 31,
+    "2": 34,
+    "3": 40,
+    "4": 39,
+    "5": 38,
+    "6": 44,
+    "7": 48,
+    "8": 43,
+    "9": 37,
+    ":": 51,
+    ";": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 71,
+    "]": 72,
+    "`": 99,
+    "a": 7,
+    "b": 24,
+    "c": 14,
+    "d": 15,
+    "e": 6,
+    "f": 21,
+    "g": 19,
+    "h": 22,
+    "i": 5,
+    "j": 42,
+    "k": 30,
+    "l": 12,
+    "m": 18,
+    "n": 9,
+    "o": 8,
+    "p": 17,
+    "q": 26,
+    "r": 11,
+    "s": 13,
+    "t": 10,
+    "u": 16,
+    "v": 20,
+    "w": 35,
+    "x": 46,
+    "y": 33,
+    "z": 23,
+    "|": 4,
+    "¥": 65,
+    "ª": 68,
+    "°": 54,
+    "²": 57,
+    "º": 69,
+    "ß": 95,
+    "à": 28,
+    "á": 58,
+    "ã": 82,
+    "ä": 79,
+    "å": 89,
+    "æ": 94,
+    "ç": 64,
+    "è": 27,
+    "é": 49,
+    "ê": 91,
+    "ë": 86,
+    "ì": 47,
+    "í": 63,
+    "î": 90,
+    "ï": 74,
+    "ð": 112,
+    "ñ": 80,
+    "ò": 36,
+    "ó": 70,
+    "ô": 88,
+    "ö": 96,
+    "ø": 81,
+    "ù": 32,
+    "ú": 73,
+    "û": 126,
+    "ü": 60,
+    "þ": 110,
+    "ÿ": 149,
+    "ā": 129,
+    "ą": 146,
+    "ć": 85,
+    "č": 76,
+    "đ": 103,
+    "ė": 104,
+    "ę": 120,
+    "ě": 102,
+    "ğ": 122,
+    "ī": 84,
+    "ı": 125,
+    "ľ": 133,
+    "ł": 67,
+    "ń": 93,
+    "ň": 107,
+    "ō": 75,
+    "ő": 109,
+    "œ": 111,
+    "ř": 100,
+    "ś": 117,
+    "ş": 97,
+    "š": 66,
+    "ū": 78,
+    "ŭ": 142,
+    "ź": 121,
+    "ż": 181,
+    "ž": 87,
+    "ș": 159,
+    "ț": 124,
+    "ə": 101,
+    "ʻ": 116,
+    "ʼ": 134,
+    "ʾ": 128,
+    "ʿ": 83,
+    "ː": 119,
+    "́": 105,
+    "̨": 172,
+    "μ": 123,
+    "а": 114,
+    "б": 113,
+    "д": 144,
+    "е": 130,
+    "л": 150,
+    "н": 145,
+    "о": 127,
+    "с": 115,
+    "у": 175,
+    "ц": 148,
+    "ѐ": 173,
+    "ё": 108,
+    "љ": 151,
+    "ң": 179,
+    "ד": 131,
+    "ה": 156,
+    "ו": 155,
+    "ة": 135,
+    "ر": 171,
+    "س": 167,
+    "ص": 168,
+    "غ": 169,
+    "ل": 166,
+    "ي": 170,
+    "ḥ": 92,
+    "ṛ": 174,
+    "ṣ": 98,
+    "ṭ": 118,
+    "ễ": 176,
+    "‐": 141,
+    "–": 77,
+    "—": 138,
+    "’": 56,
+    "′": 106,
+    "☆": 182,
+    "あ": 147,
+    "ア": 157,
+    "カ": 161,
+    "キ": 162,
+    "サ": 160,
+    "ザ": 165,
+    "ノ": 158,
+    "フ": 163,
+    "リ": 164,
+    "ン": 132,
+    "万": 154,
+    "三": 137,
+    "丰": 177,
+    "古": 152,
+    "多": 153,
+    "家": 140,
+    "峰": 178,
+    "張": 136,
+    "旅": 143,
+    "禅": 139,
+    "ꞌ": 180
+  },
+  "itl": {
+    "0": 48,
+    "1": 50,
+    "2": 43,
+    "3": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ŏ": 41,
+    "ɂ": 11,
+    "ʼ": 23,
+    "а": 7,
+    "б": 38,
+    "в": 21,
+    "г": 37,
+    "д": 36,
+    "ж": 45,
+    "з": 17,
+    "и": 9,
+    "й": 27,
+    "к": 8,
+    "л": 13,
+    "м": 16,
+    "н": 5,
+    "о": 24,
+    "п": 29,
+    "р": 30,
+    "с": 10,
+    "т": 12,
+    "у": 18,
+    "ф": 31,
+    "х": 19,
+    "ц": 39,
+    "ч": 20,
+    "ш": 49,
+    "щ": 42,
+    "ы": 26,
+    "ь": 44,
+    "э": 6,
+    "я": 47,
+    "љ": 34,
+    "њ": 32,
+    "ў": 40,
+    "ӄ": 14,
+    "ӈ": 28,
+    "ӑ": 33,
+    "ә": 22,
+    "ӽ": 25,
+    "ԓ": 15,
+    "–": 35
+  },
+  "itv": {
+    "'": 30,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 29,
+    "d": 19,
+    "e": 14,
+    "f": 23,
+    "g": 8,
+    "h": 24,
+    "i": 7,
+    "j": 26,
+    "k": 9,
+    "l": 16,
+    "m": 13,
+    "n": 6,
+    "o": 20,
+    "p": 18,
+    "q": 32,
+    "r": 15,
+    "s": 17,
+    "t": 12,
+    "u": 10,
+    "v": 27,
+    "w": 22,
+    "x": 31,
+    "y": 11,
+    "z": 25,
+    "|": 5
+  },
+  "ixl-dialect_sangasparchajul": {
+    "-": 27,
+    "0": 35,
+    "3": 34,
+    "4": 33,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 23,
+    "d": 29,
+    "e": 9,
+    "f": 30,
+    "g": 31,
+    "h": 17,
+    "i": 8,
+    "j": 24,
+    "k": 18,
+    "l": 10,
+    "m": 22,
+    "n": 12,
+    "o": 15,
+    "p": 26,
+    "q": 13,
+    "r": 28,
+    "s": 19,
+    "t": 7,
+    "u": 11,
+    "v": 16,
+    "x": 14,
+    "y": 25,
+    "z": 21,
+    "|": 4,
+    "ʼ": 6
+  },
+  "ixl-dialect_sanjuancotzal": {
+    "'": 6,
+    "-": 44,
+    "0": 31,
+    "1": 33,
+    "2": 36,
+    "3": 37,
+    "4": 34,
+    "5": 38,
+    "6": 39,
+    "7": 41,
+    "8": 42,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 19,
+    "d": 29,
+    "e": 10,
+    "f": 32,
+    "g": 30,
+    "h": 14,
+    "i": 8,
+    "j": 20,
+    "k": 15,
+    "l": 11,
+    "m": 21,
+    "n": 12,
+    "o": 13,
+    "p": 26,
+    "q": 22,
+    "r": 27,
+    "s": 16,
+    "t": 7,
+    "u": 9,
+    "v": 24,
+    "x": 17,
+    "y": 25,
+    "z": 18,
+    "|": 4,
+    "ñ": 43,
+    "–": 35,
+    "—": 28
+  },
+  "ixl-dialect_santamarianebaj": {
+    "'": 6,
+    "-": 40,
+    "0": 38,
+    "1": 39,
+    "3": 44,
+    "4": 41,
+    "5": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 21,
+    "d": 30,
+    "e": 8,
+    "f": 36,
+    "g": 31,
+    "h": 16,
+    "i": 10,
+    "j": 19,
+    "k": 13,
+    "l": 11,
+    "m": 23,
+    "n": 12,
+    "o": 17,
+    "p": 26,
+    "q": 22,
+    "r": 27,
+    "s": 14,
+    "t": 7,
+    "u": 9,
+    "v": 15,
+    "x": 18,
+    "y": 25,
+    "z": 24,
+    "|": 4,
+    "á": 34,
+    "é": 33,
+    "í": 32,
+    "ñ": 42,
+    "ó": 35,
+    "ú": 29,
+    "–": 37,
+    "—": 28
+  },
+  "izr": {
+    "'": 24,
+    "-": 37,
+    "0": 29,
+    "1": 31,
+    "2": 34,
+    "4": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 27,
+    "d": 21,
+    "e": 9,
+    "f": 20,
+    "g": 17,
+    "h": 22,
+    "i": 7,
+    "j": 30,
+    "k": 8,
+    "l": 26,
+    "m": 16,
+    "n": 6,
+    "o": 15,
+    "p": 25,
+    "r": 11,
+    "s": 14,
+    "t": 12,
+    "u": 13,
+    "v": 28,
+    "w": 19,
+    "y": 10,
+    "z": 23,
+    "|": 4,
+    "ł": 38,
+    "ŕ": 33,
+    "ٔ": 36,
+    "—": 35
+  },
+  "izz": {
+    "'": 25,
+    "-": 28,
+    "6": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 31,
+    "d": 22,
+    "e": 6,
+    "f": 32,
+    "g": 23,
+    "h": 11,
+    "i": 10,
+    "j": 29,
+    "k": 12,
+    "l": 20,
+    "m": 21,
+    "n": 8,
+    "o": 9,
+    "p": 15,
+    "r": 19,
+    "s": 26,
+    "t": 27,
+    "u": 7,
+    "v": 34,
+    "w": 24,
+    "y": 18,
+    "z": 30,
+    "|": 4,
+    "à": 43,
+    "á": 39,
+    "è": 45,
+    "é": 35,
+    "ì": 42,
+    "í": 41,
+    "ò": 46,
+    "ó": 38,
+    "ù": 37,
+    "ú": 36,
+    "ń": 48,
+    "ǹ": 47,
+    "̀": 40,
+    "́": 44,
+    "ḿ": 50,
+    "ẹ": 16,
+    "ị": 33,
+    "ọ": 17,
+    "ụ": 13
+  },
+  "jac": {
+    "'": 14,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 35,
+    "a": 5,
+    "b": 20,
+    "c": 8,
+    "d": 28,
+    "e": 10,
+    "f": 33,
+    "g": 32,
+    "h": 12,
+    "i": 6,
+    "j": 18,
+    "k": 24,
+    "l": 15,
+    "m": 17,
+    "n": 7,
+    "o": 11,
+    "p": 26,
+    "q": 31,
+    "r": 27,
+    "s": 19,
+    "t": 9,
+    "u": 16,
+    "v": 34,
+    "w": 23,
+    "x": 21,
+    "y": 13,
+    "z": 25,
+    "|": 4,
+    "̈": 22,
+    "ẍ": 29
+  },
+  "jam": {
+    "'": 31,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 28,
+    "d": 8,
+    "e": 9,
+    "f": 21,
+    "g": 19,
+    "h": 23,
+    "i": 5,
+    "j": 27,
+    "k": 17,
+    "l": 16,
+    "m": 11,
+    "n": 7,
+    "o": 12,
+    "p": 18,
+    "r": 20,
+    "s": 13,
+    "t": 14,
+    "u": 10,
+    "v": 24,
+    "w": 15,
+    "y": 25,
+    "z": 26,
+    "|": 4,
+    "—": 29
+  },
+  "jav": {
+    "!": 69,
+    "\"": 48,
+    "$": 53,
+    "%": 51,
+    "&": 55,
+    "'": 45,
+    "+": 56,
+    ",": 43,
+    "-": 26,
+    ".": 35,
+    "/": 44,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "3": 36,
+    "4": 34,
+    "5": 33,
+    "6": 38,
+    "7": 40,
+    "8": 37,
+    "9": 32,
+    ":": 47,
+    ";": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 61,
+    "]": 62,
+    "`": 72,
+    "a": 4,
+    "b": 21,
+    "c": 25,
+    "d": 18,
+    "e": 8,
+    "f": 27,
+    "g": 9,
+    "h": 20,
+    "i": 7,
+    "j": 24,
+    "k": 10,
+    "l": 15,
+    "m": 16,
+    "n": 6,
+    "o": 17,
+    "p": 19,
+    "q": 42,
+    "r": 13,
+    "s": 11,
+    "t": 14,
+    "u": 12,
+    "v": 29,
+    "w": 22,
+    "x": 41,
+    "y": 23,
+    "z": 39,
+    "|": 5,
+    "£": 63,
+    "°": 66,
+    "²": 67,
+    "á": 52,
+    "ã": 64,
+    "ç": 58,
+    "é": 46,
+    "ê": 65,
+    "í": 60,
+    "ó": 70,
+    "õ": 59,
+    "ö": 71,
+    "ü": 54,
+    "—": 57,
+    "‘": 68,
+    "”": 49
+  },
+  "jbu": {
+    "'": 30,
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 20,
+    "d": 12,
+    "e": 11,
+    "f": 25,
+    "g": 26,
+    "h": 24,
+    "i": 7,
+    "j": 21,
+    "k": 10,
+    "l": 32,
+    "m": 18,
+    "n": 6,
+    "o": 15,
+    "p": 23,
+    "q": 49,
+    "r": 16,
+    "s": 17,
+    "t": 19,
+    "u": 8,
+    "v": 28,
+    "w": 13,
+    "y": 14,
+    "z": 22,
+    "|": 4,
+    "­": 29,
+    "à": 27,
+    "á": 35,
+    "â": 50,
+    "è": 39,
+    "é": 41,
+    "ì": 40,
+    "í": 31,
+    "ò": 34,
+    "ó": 46,
+    "ù": 38,
+    "ú": 36,
+    "ā": 37,
+    "ē": 43,
+    "ī": 33,
+    "ō": 42,
+    "ū": 48,
+    "̀": 45,
+    "ḿ": 47
+  },
+  "jen": {
+    "'": 38,
+    "8": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 11,
+    "c": 30,
+    "d": 24,
+    "e": 10,
+    "f": 22,
+    "g": 16,
+    "h": 21,
+    "i": 9,
+    "j": 35,
+    "k": 18,
+    "l": 17,
+    "m": 19,
+    "n": 5,
+    "o": 20,
+    "p": 25,
+    "r": 34,
+    "s": 12,
+    "t": 14,
+    "u": 23,
+    "v": 26,
+    "w": 13,
+    "y": 15,
+    "z": 29,
+    "|": 4,
+    "ã": 31,
+    "ĩ": 36,
+    "ũ": 32,
+    "ɔ": 33,
+    "ə": 7,
+    "ɛ": 27,
+    "ɨ": 6,
+    "̃": 28
+  },
+  "jic": {
+    "'": 23,
+    "0": 42,
+    "1": 43,
+    "2": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 17,
+    "d": 24,
+    "e": 12,
+    "f": 38,
+    "g": 33,
+    "h": 39,
+    "i": 11,
+    "j": 6,
+    "k": 45,
+    "l": 10,
+    "m": 16,
+    "n": 7,
+    "o": 15,
+    "p": 9,
+    "q": 26,
+    "r": 28,
+    "s": 8,
+    "t": 13,
+    "u": 14,
+    "v": 25,
+    "w": 21,
+    "x": 41,
+    "y": 18,
+    "z": 37,
+    "|": 4,
+    "á": 22,
+    "é": 27,
+    "í": 29,
+    "ñ": 40,
+    "ó": 30,
+    "ú": 31,
+    "ü": 19,
+    "ǘ": 36,
+    "̈": 32,
+    "—": 35,
+    "ꞌ": 20
+  },
+  "jiv": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 20,
+    "e": 15,
+    "h": 14,
+    "i": 6,
+    "j": 17,
+    "k": 13,
+    "m": 12,
+    "n": 8,
+    "p": 22,
+    "r": 10,
+    "s": 11,
+    "t": 9,
+    "u": 7,
+    "w": 21,
+    "y": 18,
+    "|": 5,
+    "á": 19,
+    "é": 26,
+    "í": 23,
+    "ð": 27,
+    "ú": 25,
+    "̱": 16,
+    "ṉ": 24
+  },
+  "jmc": {
+    "'": 27,
+    "-": 25,
+    "0": 29,
+    "1": 32,
+    "2": 35,
+    "3": 34,
+    "4": 30,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "d": 19,
+    "e": 8,
+    "f": 21,
+    "g": 24,
+    "h": 20,
+    "i": 6,
+    "j": 31,
+    "k": 10,
+    "l": 15,
+    "m": 13,
+    "n": 7,
+    "o": 12,
+    "p": 26,
+    "r": 18,
+    "s": 14,
+    "t": 22,
+    "u": 11,
+    "v": 17,
+    "w": 16,
+    "y": 9,
+    "z": 28,
+    "|": 5
+  },
+  "jmd": {
+    "'": 29,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 27,
+    "d": 23,
+    "e": 7,
+    "f": 18,
+    "g": 21,
+    "h": 24,
+    "i": 10,
+    "j": 26,
+    "k": 15,
+    "l": 13,
+    "m": 9,
+    "n": 6,
+    "o": 14,
+    "p": 22,
+    "r": 8,
+    "s": 12,
+    "t": 11,
+    "u": 16,
+    "w": 20,
+    "y": 17,
+    "z": 28,
+    "|": 4
+  },
+  "jpn": {
+    "%": 661,
+    "&": 1702,
+    "'": 1669,
+    "+": 1640,
+    ",": 393,
+    "-": 943,
+    ".": 673,
+    "/": 1054,
+    "0": 39,
+    "1": 40,
+    "2": 57,
+    "3": 110,
+    "4": 145,
+    "5": 121,
+    "6": 151,
+    "7": 197,
+    "8": 153,
+    "9": 118,
+    ":": 1381,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 147,
+    "b": 610,
+    "c": 317,
+    "d": 340,
+    "e": 293,
+    "f": 555,
+    "g": 334,
+    "h": 578,
+    "i": 360,
+    "j": 903,
+    "k": 416,
+    "l": 378,
+    "m": 234,
+    "n": 332,
+    "o": 303,
+    "p": 229,
+    "q": 1526,
+    "r": 341,
+    "s": 338,
+    "t": 364,
+    "u": 481,
+    "v": 655,
+    "w": 1014,
+    "x": 1095,
+    "y": 896,
+    "z": 1323,
+    "²": 1612,
+    "×": 1471,
+    "–": 2046,
+    "—": 1996,
+    "℃": 1449,
+    "、": 241,
+    "々": 237,
+    "「": 407,
+    "」": 397,
+    "〜": 2143,
+    "ぁ": 2026,
+    "あ": 28,
+    "い": 14,
+    "う": 31,
+    "ぇ": 2087,
+    "え": 69,
+    "お": 152,
+    "か": 25,
+    "が": 16,
+    "き": 41,
+    "ぎ": 519,
+    "く": 37,
+    "ぐ": 524,
+    "け": 47,
+    "げ": 216,
+    "こ": 23,
+    "ご": 371,
+    "さ": 30,
+    "ざ": 568,
+    "し": 6,
+    "じ": 208,
+    "す": 11,
+    "ず": 161,
+    "せ": 60,
+    "ぜ": 712,
+    "そ": 53,
+    "ぞ": 876,
+    "た": 9,
+    "だ": 86,
+    "ち": 76,
+    "っ": 33,
+    "つ": 59,
+    "づ": 516,
+    "て": 17,
+    "で": 8,
+    "と": 13,
+    "ど": 64,
+    "な": 20,
+    "に": 5,
+    "ぬ": 1773,
+    "ね": 1083,
+    "の": 4,
+    "は": 7,
+    "ば": 84,
+    "ぱ": 1635,
+    "ひ": 1114,
+    "び": 311,
+    "ぴ": 2261,
+    "ふ": 1050,
+    "ぶ": 570,
+    "ぷ": 1523,
+    "へ": 373,
+    "べ": 124,
+    "ぺ": 1829,
+    "ほ": 125,
+    "ぼ": 649,
+    "ぽ": 1444,
+    "ま": 10,
+    "み": 129,
+    "む": 322,
+    "め": 50,
+    "も": 29,
+    "ゃ": 1140,
+    "や": 49,
+    "ゅ": 2264,
+    "ゆ": 760,
+    "ょ": 142,
+    "よ": 42,
+    "ら": 27,
+    "り": 22,
+    "る": 15,
+    "れ": 21,
+    "ろ": 264,
+    "わ": 74,
+    "を": 12,
+    "ん": 46,
+    "ァ": 249,
+    "ア": 36,
+    "ィ": 94,
+    "イ": 35,
+    "ゥ": 565,
+    "ウ": 114,
+    "ェ": 134,
+    "エ": 120,
+    "ォ": 402,
+    "オ": 72,
+    "カ": 61,
+    "ガ": 192,
+    "キ": 103,
+    "ギ": 218,
+    "ク": 43,
+    "グ": 96,
+    "ケ": 179,
+    "ゲ": 427,
+    "コ": 80,
+    "ゴ": 250,
+    "サ": 92,
+    "ザ": 230,
+    "シ": 58,
+    "ジ": 63,
+    "ス": 24,
+    "ズ": 112,
+    "セ": 159,
+    "ゼ": 894,
+    "ソ": 193,
+    "ゾ": 1041,
+    "タ": 56,
+    "ダ": 126,
+    "チ": 95,
+    "ッ": 44,
+    "ツ": 166,
+    "ヅ": 2211,
+    "テ": 70,
+    "デ": 139,
+    "ト": 32,
+    "ド": 48,
+    "ナ": 148,
+    "ニ": 89,
+    "ヌ": 950,
+    "ネ": 156,
+    "ノ": 213,
+    "ハ": 169,
+    "バ": 81,
+    "パ": 128,
+    "ヒ": 450,
+    "ビ": 132,
+    "ピ": 146,
+    "フ": 54,
+    "ブ": 82,
+    "プ": 88,
+    "ヘ": 463,
+    "ベ": 207,
+    "ペ": 260,
+    "ホ": 271,
+    "ボ": 224,
+    "ポ": 174,
+    "マ": 87,
+    "ミ": 163,
+    "ム": 105,
+    "メ": 115,
+    "モ": 288,
+    "ャ": 99,
+    "ヤ": 368,
+    "ュ": 130,
+    "ユ": 488,
+    "ョ": 181,
+    "ヨ": 445,
+    "ラ": 34,
+    "リ": 38,
+    "ル": 26,
+    "レ": 68,
+    "ロ": 55,
+    "ワ": 291,
+    "ン": 19,
+    "ヴ": 405,
+    "ヶ": 1881,
+    "・": 777,
+    "ー": 18,
+    "一": 93,
+    "丁": 2157,
+    "七": 1726,
+    "万": 438,
+    "丈": 2254,
+    "三": 805,
+    "上": 91,
+    "下": 236,
+    "不": 365,
+    "与": 532,
+    "世": 155,
+    "丘": 1251,
+    "両": 447,
+    "並": 865,
+    "中": 90,
+    "丸": 1371,
+    "丹": 2059,
+    "主": 238,
+    "久": 1377,
+    "乏": 1855,
+    "乗": 417,
+    "乙": 2098,
+    "九": 1232,
+    "乞": 1568,
+    "乱": 1790,
+    "乳": 1380,
+    "乾": 1333,
+    "亀": 1503,
+    "了": 847,
+    "予": 200,
+    "争": 471,
+    "事": 149,
+    "二": 581,
+    "互": 1101,
+    "五": 1806,
+    "井": 1509,
+    "些": 2032,
+    "亜": 975,
+    "亡": 502,
+    "交": 375,
+    "京": 913,
+    "人": 45,
+    "仁": 1520,
+    "今": 348,
+    "介": 1249,
+    "仏": 1550,
+    "仕": 883,
+    "他": 228,
+    "付": 327,
+    "代": 223,
+    "令": 1394,
+    "以": 175,
+    "仮": 1383,
+    "仰": 1344,
+    "仲": 1224,
+    "件": 353,
+    "任": 464,
+    "企": 963,
+    "伏": 1739,
+    "伐": 2233,
+    "休": 1680,
+    "会": 137,
+    "伝": 469,
+    "伯": 1895,
+    "伴": 1136,
+    "伸": 992,
+    "似": 1004,
+    "佇": 1621,
+    "位": 333,
+    "低": 399,
+    "住": 316,
+    "佐": 1266,
+    "体": 144,
+    "何": 269,
+    "余": 774,
+    "作": 215,
+    "併": 1544,
+    "使": 211,
+    "例": 474,
+    "供": 275,
+    "依": 902,
+    "価": 430,
+    "侵": 715,
+    "侶": 1522,
+    "便": 556,
+    "係": 455,
+    "促": 1469,
+    "俗": 1539,
+    "保": 268,
+    "信": 467,
+    "俣": 2102,
+    "修": 960,
+    "俳": 1809,
+    "俺": 1606,
+    "倉": 1973,
+    "個": 517,
+    "倍": 856,
+    "倒": 796,
+    "候": 620,
+    "借": 1240,
+    "倣": 1979,
+    "値": 678,
+    "倫": 1611,
+    "偉": 2094,
+    "偏": 2153,
+    "停": 1281,
+    "健": 972,
+    "側": 668,
+    "偵": 1714,
+    "偶": 1851,
+    "偽": 1686,
+    "傑": 1414,
+    "傘": 1671,
+    "備": 477,
+    "催": 520,
+    "債": 1628,
+    "傷": 503,
+    "傾": 630,
+    "僅": 1967,
+    "働": 625,
+    "像": 616,
+    "僕": 1774,
+    "僚": 1757,
+    "僧": 1521,
+    "儀": 948,
+    "億": 835,
+    "償": 1954,
+    "優": 415,
+    "儲": 1682,
+    "元": 468,
+    "兄": 1770,
+    "充": 1713,
+    "兆": 1241,
+    "先": 227,
+    "光": 190,
+    "克": 1499,
+    "免": 1055,
+    "児": 731,
+    "党": 886,
+    "入": 113,
+    "全": 214,
+    "八": 1451,
+    "公": 337,
+    "六": 1850,
+    "共": 521,
+    "兵": 766,
+    "其": 2147,
+    "具": 672,
+    "典": 974,
+    "兼": 1810,
+    "内": 222,
+    "円": 1167,
+    "冊": 1750,
+    "再": 508,
+    "冑": 2216,
+    "冒": 1308,
+    "冗": 1835,
+    "写": 483,
+    "冠": 1402,
+    "冬": 820,
+    "冶": 2132,
+    "冷": 1139,
+    "准": 1036,
+    "凌": 1701,
+    "凍": 1805,
+    "凝": 1759,
+    "処": 907,
+    "出": 100,
+    "函": 1476,
+    "刀": 2230,
+    "刃": 2064,
+    "分": 111,
+    "切": 597,
+    "刈": 1502,
+    "刊": 1861,
+    "刑": 636,
+    "列": 1094,
+    "初": 257,
+    "判": 343,
+    "別": 329,
+    "利": 226,
+    "到": 999,
+    "制": 299,
+    "刷": 1045,
+    "券": 808,
+    "刺": 930,
+    "刻": 802,
+    "剃": 1933,
+    "則": 922,
+    "削": 1129,
+    "前": 140,
+    "剣": 1583,
+    "剤": 1391,
+    "剥": 2141,
+    "副": 1564,
+    "剰": 1458,
+    "割": 446,
+    "創": 1173,
+    "劇": 1170,
+    "力": 172,
+    "功": 721,
+    "加": 318,
+    "劣": 2263,
+    "助": 724,
+    "努": 1118,
+    "励": 1596,
+    "労": 997,
+    "効": 643,
+    "勃": 1984,
+    "勅": 2162,
+    "勇": 1623,
+    "勉": 1409,
+    "動": 106,
+    "勘": 1715,
+    "務": 352,
+    "勝": 485,
+    "募": 1398,
+    "勢": 771,
+    "勤": 1537,
+    "勧": 945,
+    "勲": 2016,
+    "勾": 1364,
+    "匂": 1278,
+    "包": 1487,
+    "化": 136,
+    "北": 309,
+    "匹": 1275,
+    "区": 567,
+    "医": 551,
+    "匿": 1907,
+    "十": 476,
+    "千": 654,
+    "午": 726,
+    "半": 542,
+    "卒": 1133,
+    "卓": 1700,
+    "協": 656,
+    "南": 346,
+    "単": 461,
+    "博": 1003,
+    "占": 1086,
+    "印": 703,
+    "危": 539,
+    "即": 1799,
+    "却": 1783,
+    "卵": 1710,
+    "卿": 1374,
+    "厄": 2125,
+    "厚": 1169,
+    "原": 232,
+    "厨": 2206,
+    "厳": 869,
+    "去": 554,
+    "参": 538,
+    "又": 2232,
+    "及": 888,
+    "友": 933,
+    "双": 1513,
+    "反": 629,
+    "収": 462,
+    "叔": 1921,
+    "取": 225,
+    "受": 182,
+    "叙": 2015,
+    "口": 658,
+    "古": 600,
+    "句": 1601,
+    "叩": 1152,
+    "只": 2241,
+    "叫": 1548,
+    "召": 1871,
+    "可": 171,
+    "台": 534,
+    "叱": 2208,
+    "史": 396,
+    "右": 801,
+    "叶": 1607,
+    "号": 431,
+    "司": 741,
+    "各": 719,
+    "合": 79,
+    "��": 2091,
+    "吊": 1668,
+    "同": 185,
+    "名": 209,
+    "后": 2234,
+    "吐": 1650,
+    "向": 210,
+    "君": 1685,
+    "否": 827,
+    "含": 746,
+    "吸": 976,
+    "吹": 1305,
+    "呂": 1966,
+    "呆": 2204,
+    "呈": 2193,
+    "告": 261,
+    "周": 507,
+    "呪": 1589,
+    "味": 363,
+    "呼": 302,
+    "命": 566,
+    "和": 618,
+    "品": 487,
+    "員": 289,
+    "哨": 1719,
+    "哲": 1345,
+    "哺": 1379,
+    "唆": 884,
+    "唇": 1887,
+    "唯": 813,
+    "唱": 1599,
+    "唸": 1507,
+    "商": 705,
+    "問": 313,
+    "啓": 1603,
+    "善": 1000,
+    "喉": 1467,
+    "喜": 1203,
+    "喩": 2248,
+    "喪": 2092,
+    "営": 761,
+    "嗅": 2136,
+    "嘆": 1927,
+    "嘔": 1649,
+    "嘘": 1301,
+    "噂": 2181,
+    "噛": 1486,
+    "器": 444,
+    "噴": 1510,
+    "囚": 1373,
+    "四": 1295,
+    "回": 239,
+    "因": 439,
+    "団": 537,
+    "困": 1243,
+    "囲": 518,
+    "図": 785,
+    "固": 1078,
+    "国": 52,
+    "圏": 1146,
+    "園": 682,
+    "土": 776,
+    "圧": 634,
+    "在": 189,
+    "地": 62,
+    "坂": 2182,
+    "均": 1868,
+    "坦": 1734,
+    "垂": 1904,
+    "型": 460,
+    "垣": 1694,
+    "埋": 1039,
+    "城": 864,
+    "域": 242,
+    "執": 1563,
+    "培": 2011,
+    "基": 409,
+    "堀": 2120,
+    "堂": 1721,
+    "堆": 1890,
+    "堕": 2167,
+    "堤": 1393,
+    "堪": 1659,
+    "報": 162,
+    "場": 67,
+    "塀": 2050,
+    "塊": 1614,
+    "塔": 1260,
+    "塗": 1327,
+    "塞": 1105,
+    "塩": 1318,
+    "填": 1992,
+    "塹": 2095,
+    "境": 596,
+    "墓": 892,
+    "増": 583,
+    "墜": 929,
+    "墟": 1673,
+    "壁": 1236,
+    "壇": 1330,
+    "壊": 816,
+    "壌": 1749,
+    "壕": 2096,
+    "士": 505,
+    "声": 711,
+    "売": 912,
+    "変": 276,
+    "夏": 688,
+    "夕": 1454,
+    "外": 183,
+    "多": 117,
+    "夜": 552,
+    "夢": 1074,
+    "大": 66,
+    "天": 619,
+    "太": 595,
+    "夫": 915,
+    "央": 890,
+    "失": 587,
+    "奇": 1797,
+    "奈": 1937,
+    "奉": 1961,
+    "奏": 1213,
+    "契": 1084,
+    "奥": 1993,
+    "奨": 1873,
+    "奪": 1051,
+    "奮": 1577,
+    "女": 394,
+    "奴": 1807,
+    "好": 762,
+    "如": 1796,
+    "妃": 1267,
+    "妊": 1911,
+    "妖": 1938,
+    "妙": 1481,
+    "妥": 2018,
+    "妨": 1294,
+    "妹": 1533,
+    "妻": 1304,
+    "姉": 1532,
+    "始": 273,
+    "姓": 1878,
+    "委": 713,
+    "姦": 1565,
+    "姫": 1328,
+    "姿": 994,
+    "威": 798,
+    "娘": 1666,
+    "娠": 1912,
+    "娯": 1446,
+    "娼": 2133,
+    "婚": 845,
+    "婦": 1038,
+    "媒": 1448,
+    "媽": 2103,
+    "嫁": 2044,
+    "嫌": 1437,
+    "嬉": 1964,
+    "子": 141,
+    "孔": 1196,
+    "字": 714,
+    "存": 390,
+    "季": 561,
+    "孤": 1408,
+    "学": 98,
+    "孫": 1559,
+    "孵": 2070,
+    "宅": 717,
+    "宇": 579,
+    "守": 793,
+    "安": 286,
+    "宋": 1883,
+    "完": 877,
+    "宗": 718,
+    "官": 404,
+    "宙": 580,
+    "定": 122,
+    "宛": 1399,
+    "宜": 2245,
+    "宝": 1886,
+    "実": 265,
+    "客": 354,
+    "宣": 1480,
+    "室": 1025,
+    "宮": 1268,
+    "害": 381,
+    "宴": 1588,
+    "家": 220,
+    "容": 657,
+    "宿": 498,
+    "寂": 1442,
+    "寄": 775,
+    "密": 747,
+    "富": 914,
+    "寒": 812,
+    "寛": 2038,
+    "寝": 958,
+    "察": 366,
+    "寡": 2189,
+    "寧": 2061,
+    "審": 1077,
+    "寮": 1630,
+    "寵": 2259,
+    "寸": 1704,
+    "寺": 839,
+    "対": 320,
+    "寿": 1592,
+    "封": 1285,
+    "専": 744,
+    "射": 675,
+    "将": 979,
+    "尉": 2014,
+    "尊": 1075,
+    "尋": 1832,
+    "導": 617,
+    "小": 285,
+    "少": 389,
+    "尖": 1401,
+    "尤": 2255,
+    "就": 729,
+    "尻": 2170,
+    "尽": 1580,
+    "尾": 1396,
+    "尿": 962,
+    "局": 425,
+    "居": 919,
+    "屈": 1801,
+    "届": 848,
+    "屋": 690,
+    "展": 1029,
+    "属": 541,
+    "屠": 1716,
+    "層": 861,
+    "履": 1420,
+    "屯": 1632,
+    "山": 710,
+    "岐": 1795,
+    "岩": 1148,
+    "岳": 2058,
+    "岸": 716,
+    "峡": 1732,
+    "島": 270,
+    "峻": 1733,
+    "崇": 1755,
+    "崎": 1264,
+    "崔": 2135,
+    "崖": 1466,
+    "崩": 1009,
+    "嵐": 1376,
+    "川": 614,
+    "州": 290,
+    "巡": 1175,
+    "巣": 1441,
+    "工": 787,
+    "左": 942,
+    "巧": 1879,
+    "巨": 1115,
+    "差": 843,
+    "己": 1187,
+    "巻": 831,
+    "市": 280,
+    "布": 1171,
+    "希": 957,
+    "帝": 1010,
+    "帥": 1660,
+    "師": 842,
+    "席": 983,
+    "帯": 486,
+    "帰": 531,
+    "帳": 1676,
+    "常": 150,
+    "帽": 1792,
+    "幅": 674,
+    "幕": 1357,
+    "幣": 1901,
+    "干": 1147,
+    "平": 644,
+    "年": 65,
+    "幸": 1016,
+    "幹": 1052,
+    "幻": 1763,
+    "幼": 1731,
+    "幾": 1670,
+    "庁": 1436,
+    "広": 388,
+    "床": 1691,
+    "序": 1185,
+    "底": 926,
+    "店": 562,
+    "府": 418,
+    "度": 160,
+    "���": 1128,
+    "庫": 1546,
+    "庭": 621,
+    "庵": 2097,
+    "康": 1012,
+    "廃": 1416,
+    "廊": 1648,
+    "廟": 2122,
+    "延": 853,
+    "廷": 1296,
+    "建": 420,
+    "弁": 1018,
+    "弊": 1665,
+    "式": 362,
+    "弓": 852,
+    "引": 434,
+    "弟": 1361,
+    "弦": 1956,
+    "弧": 2224,
+    "弩": 2219,
+    "弱": 1137,
+    "張": 527,
+    "強": 421,
+    "弾": 1073,
+    "当": 186,
+    "彗": 1478,
+    "彙": 1698,
+    "形": 315,
+    "彩": 1433,
+    "彫": 1738,
+    "彰": 2112,
+    "影": 336,
+    "彷": 1707,
+    "役": 602,
+    "彼": 176,
+    "彿": 1708,
+    "往": 2220,
+    "征": 1071,
+    "径": 2000,
+    "待": 635,
+    "律": 651,
+    "後": 131,
+    "徐": 1491,
+    "徒": 652,
+    "従": 748,
+    "得": 307,
+    "御": 1688,
+    "復": 594,
+    "循": 1209,
+    "微": 743,
+    "徴": 1011,
+    "心": 342,
+    "必": 165,
+    "忍": 1624,
+    "志": 1280,
+    "忘": 1574,
+    "忙": 1819,
+    "応": 443,
+    "忠": 2111,
+    "快": 977,
+    "念": 400,
+    "怒": 1362,
+    "怖": 862,
+    "思": 355,
+    "怠": 1631,
+    "急": 670,
+    "性": 109,
+    "怨": 2265,
+    "怪": 1027,
+    "怯": 1455,
+    "恋": 1791,
+    "恍": 1753,
+    "恐": 533,
+    "恒": 1231,
+    "恥": 1725,
+    "恨": 2041,
+    "恩": 1955,
+    "息": 838,
+    "恰": 2217,
+    "恵": 1885,
+    "悔": 1922,
+    "患": 928,
+    "悩": 1506,
+    "悪": 586,
+    "悲": 1777,
+    "情": 428,
+    "惑": 939,
+    "惚": 1754,
+    "惜": 2260,
+    "惟": 1920,
+    "惨": 1988,
+    "想": 496,
+    "惹": 1991,
+    "愉": 1948,
+    "意": 245,
+    "愕": 2210,
+    "愚": 2045,
+    "愛": 1197,
+    "感": 344,
+    "慈": 1825,
+    "態": 386,
+    "慌": 2205,
+    "慎": 1477,
+    "慢": 1618,
+    "慣": 723,
+    "慮": 1479,
+    "憎": 2115,
+    "憑": 1767,
+    "憧": 2146,
+    "憩": 2124,
+    "憲": 984,
+    "憶": 1134,
+    "憾": 1508,
+    "懇": 1929,
+    "懐": 1766,
+    "懲": 1456,
+    "懸": 882,
+    "成": 233,
+    "我": 906,
+    "戒": 1270,
+    "或": 2258,
+    "戚": 1202,
+    "戦": 184,
+    "戯": 1549,
+    "戴": 2127,
+    "戸": 1866,
+    "戻": 574,
+    "房": 1918,
+    "所": 154,
+    "扇": 1474,
+    "扉": 1483,
+    "手": 104,
+    "才": 1785,
+    "打": 923,
+    "払": 991,
+    "扮": 2187,
+    "扱": 720,
+    "批": 951,
+    "承": 987,
+    "技": 742,
+    "把": 1256,
+    "抑": 1061,
+    "投": 528,
+    "抗": 495,
+    "折": 859,
+    "抜": 1090,
+    "択": 837,
+    "披": 969,
+    "抱": 1440,
+    "抵": 1555,
+    "抹": 2025,
+    "押": 1098,
+    "抽": 1781,
+    "担": 795,
+    "拍": 2107,
+    "拐": 1968,
+    "拒": 973,
+    "拓": 1385,
+    "拘": 1341,
+    "招": 1253,
+    "拝": 1756,
+    "拠": 608,
+    "拡": 1031,
+    "括": 1985,
+    "拭": 2178,
+    "拶": 1987,
+    "拷": 1997,
+    "拾": 1582,
+    "持": 247,
+    "指": 437,
+    "挑": 1836,
+    "挙": 413,
+    "挟": 1587,
+    "挨": 1986,
+    "挫": 1683,
+    "振": 826,
+    "挿": 1827,
+    "捉": 1615,
+    "捕": 639,
+    "捜": 828,
+    "捧": 1501,
+    "捨": 1457,
+    "掃": 1276,
+    "授": 898,
+    "掌": 1488,
+    "排": 1349,
+    "掘": 1126,
+    "掛": 1432,
+    "採": 1067,
+    "探": 693,
+    "接": 573,
+    "控": 1687,
+    "推": 804,
+    "措": 1064,
+    "掲": 1225,
+    "掴": 2037,
+    "掻": 1705,
+    "揃": 1274,
+    "描": 1255,
+    "提": 369,
+    "揚": 1818,
+    "換": 810,
+    "握": 1066,
+    "揮": 968,
+    "援": 765,
+    "揺": 1849,
+    "損": 860,
+    "搬": 1217,
+    "搭": 1347,
+    "携": 878,
+    "搾": 1405,
+    "摂": 1307,
+    "摘": 1303,
+    "摩": 2021,
+    "撃": 424,
+    "撒": 1744,
+    "撤": 1800,
+    "撮": 694,
+    "撲": 1724,
+    "擁": 1902,
+    "操": 871,
+    "擦": 1919,
+    "擬": 1527,
+    "支": 511,
+    "改": 755,
+    "攻": 576,
+    "放": 331,
+    "政": 326,
+    "故": 572,
+    "敏": 2029,
+    "救": 900,
+    "敗": 887,
+    "教": 212,
+    "敢": 1999,
+    "散": 830,
+    "敬": 1425,
+    "数": 158,
+    "整": 901,
+    "敵": 946,
+    "敷": 873,
+    "文": 205,
+    "斉": 1519,
+    "斎": 1338,
+    "斑": 2161,
+    "斗": 1646,
+    "料": 310,
+    "斜": 1226,
+    "斧": 1931,
+    "断": 480,
+    "斯": 2252,
+    "新": 328,
+    "方": 133,
+    "施": 604,
+    "旅": 180,
+    "旋": 1426,
+    "族": 300,
+    "旗": 1189,
+    "既": 1421,
+    "日": 78,
+    "旧": 1103,
+    "早": 940,
+    "旬": 1131,
+    "昆": 1865,
+    "昇": 1339,
+    "昌": 2075,
+    "明": 219,
+    "昏": 2169,
+    "易": 1044,
+    "昔": 1329,
+    "星": 408,
+    "映": 525,
+    "春": 1287,
+    "昧": 1321,
+    "昨": 833,
+    "是": 1343,
+    "昼": 1322,
+    "時": 101,
+    "晒": 2209,
+    "晩": 1113,
+    "普": 1404,
+    "景": 832,
+    "晰": 1397,
+    "晴": 1424,
+    "晶": 1517,
+    "暇": 1681,
+    "暑": 1021,
+    "暖": 1109,
+    "暗": 970,
+    "暢": 2093,
+    "暫": 1678,
+    "暮": 1206,
+    "暴": 575,
+    "曖": 1320,
+    "曜": 395,
+    "曲": 857,
+    "曳": 2213,
+    "更": 908,
+    "書": 246,
+    "曹": 1495,
+    "替": 1453,
+    "最": 107,
+    "月": 143,
+    "有": 248,
+    "服": 899,
+    "朗": 1368,
+    "望": 624,
+    "朝": 683,
+    "期": 254,
+    "朦": 2256,
+    "朧": 2257,
+    "木": 811,
+    "未": 863,
+    "末": 840,
+    "本": 164,
+    "札": 1490,
+    "机": 1847,
+    "朽": 2047,
+    "杉": 1909,
+    "材": 751,
+    "村": 937,
+    "杖": 2231,
+    "束": 1418,
+    "条": 817,
+    "来": 312,
+    "杭": 1656,
+    "杯": 1439,
+    "東": 466,
+    "松": 2009,
+    "板": 1001,
+    "析": 1787,
+    "枕": 2030,
+    "林": 995,
+    "枚": 986,
+    "果": 358,
+    "枝": 1485,
+    "枠": 1411,
+    "枢": 1896,
+    "枯": 1949,
+    "架": 1775,
+    "柄": 1309,
+    "染": 423,
+    "柔": 1069,
+    "柱": 1450,
+    "柳": 2049,
+    "柵": 1655,
+    "査": 339,
+    "栄": 1205,
+    "栓": 1915,
+    "校": 822,
+    "株": 1644,
+    "核": 1030,
+    "根": 1022,
+    "格": 666,
+    "栽": 2010,
+    "案": 866,
+    "桜": 1514,
+    "桟": 1892,
+    "梅": 1633,
+    "梨": 2168,
+    "械": 1802,
+    "棄": 1538,
+    "棒": 1388,
+    "棚": 1803,
+    "棟": 1897,
+    "森": 1284,
+    "棺": 1493,
+    "椅": 1314,
+    "植": 677,
+    "椎": 2083,
+    "検": 522,
+    "楕": 2221,
+    "業": 231,
+    "極": 601,
+    "楼": 1765,
+    "楽": 398,
+    "概": 949,
+    "榜": 1545,
+    "構": 540,
+    "槍": 1706,
+    "様": 457,
+    "槽": 2022,
+    "標": 676,
+    "模": 650,
+    "権": 510,
+    "横": 947,
+    "樹": 1823,
+    "橄": 2139,
+    "橋": 910,
+    "橙": 2156,
+    "機": 194,
+    "欄": 1661,
+    "欖": 2140,
+    "欠": 701,
+    "次": 545,
+    "欧": 645,
+    "欲": 1237,
+    "欺": 1617,
+    "歌": 788,
+    "歓": 1853,
+    "止": 449,
+    "正": 367,
+    "武": 1210,
+    "歩": 489,
+    "歯": 1151,
+    "歳": 663,
+    "歴": 473,
+    "死": 359,
+    "殉": 2090,
+    "殊": 1272,
+    "残": 546,
+    "殖": 1172,
+    "殴": 2236,
+    "段": 905,
+    "殺": 1024,
+    "殻": 1711,
+    "殿": 806,
+    "毅": 2174,
+    "母": 909,
+    "毎": 790,
+    "毒": 1088,
+    "比": 584,
+    "毛": 1551,
+    "氏": 198,
+    "民": 325,
+    "気": 188,
+    "水": 168,
+    "氷": 706,
+    "永": 1431,
+    "汁": 1925,
+    "求": 523,
+    "汎": 2229,
+    "汗": 2020,
+    "汚": 1019,
+    "汝": 2253,
+    "江": 1965,
+    "池": 1860,
+    "汰": 2072,
+    "汲": 2067,
+    "決": 277,
+    "沃": 1994,
+    "沈": 1452,
+    "沖": 1634,
+    "没": 1658,
+    "沢": 1193,
+    "河": 789,
+    "沸": 2118,
+    "油": 880,
+    "治": 615,
+    "沼": 2078,
+    "沿": 784,
+    "況": 671,
+    "泉": 1930,
+    "泊": 559,
+    "泌": 2035,
+    "法": 135,
+    "泡": 1315,
+    "波": 543,
+    "泣": 1547,
+    "泥": 1626,
+    "注": 735,
+    "泳": 783,
+    "洋": 515,
+    "洗": 1419,
+    "洞": 779,
+    "津": 1028,
+    "洪": 1096,
+    "活": 323,
+    "派": 953,
+    "流": 305,
+    "浄": 1736,
+    "浅": 1613,
+    "浪": 1642,
+    "浮": 1037,
+    "浴": 1594,
+    "海": 274,
+    "浸": 1664,
+    "消": 627,
+    "涙": 1586,
+    "涯": 1654,
+    "液": 870,
+    "涼": 1412,
+    "淘": 2071,
+    "淡": 1935,
+    "深": 680,
+    "混": 1117,
+    "淹": 1679,
+    "添": 1312,
+    "清": 1035,
+    "渇": 2040,
+    "済": 547,
+    "渉": 1168,
+    "渋": 2052,
+    "渓": 1123,
+    "減": 803,
+    "渡": 550,
+    "渦": 1647,
+    "温": 738,
+    "測": 778,
+    "港": 648,
+    "湖": 1089,
+    "湾": 1056,
+    "湿": 1244,
+    "満": 752,
+    "源": 642,
+    "準": 440,
+    "溜": 1924,
+    "溢": 1370,
+    "溶": 1914,
+    "溺": 2036,
+    "滅": 1286,
+    "滑": 849,
+    "滝": 917,
+    "滞": 631,
+    "滴": 2185,
+    "漁": 1423,
+    "漂": 1936,
+    "漏": 875,
+    "漑": 2100,
+    "演": 609,
+    "漕": 1941,
+    "漠": 1518,
+    "漢": 1882,
+    "漫": 1833,
+    "漬": 2031,
+    "潔": 1464,
+    "潜": 512,
+    "潤": 1662,
+    "潮": 1297,
+    "澄": 2267,
+    "激": 606,
+    "濁": 2076,
+    "濃": 1087,
+    "濡": 1629,
+    "濤": 1352,
+    "濯": 1554,
+    "瀕": 2154,
+    "瀬": 1963,
+    "灌": 2099,
+    "火": 465,
+    "灯": 1581,
+    "灰": 1780,
+    "災": 1235,
+    "炉": 1153,
+    "炎": 1761,
+    "炙": 2138,
+    "炭": 1192,
+    "点": 251,
+    "為": 730,
+    "烈": 1856,
+    "焉": 1489,
+    "無": 475,
+    "焦": 753,
+    "然": 494,
+    "焼": 1257,
+    "煉": 2152,
+    "煙": 1917,
+    "照": 867,
+    "煮": 1839,
+    "煽": 1720,
+    "熊": 1201,
+    "熟": 1258,
+    "熱": 479,
+    "熾": 2134,
+    "燃": 815,
+    "燕": 2053,
+    "燥": 1815,
+    "爆": 535,
+    "爪": 1082,
+    "爬": 2003,
+    "爵": 1834,
+    "父": 1324,
+    "爽": 2069,
+    "片": 1057,
+    "版": 868,
+    "牛": 1639,
+    "牝": 2202,
+    "牡": 2056,
+    "牧": 1190,
+    "物": 73,
+    "牲": 1562,
+    "特": 259,
+    "犀": 1977,
+    "犠": 1561,
+    "犬": 1291,
+    "犯": 1032,
+    "状": 240,
+    "狂": 1672,
+    "狙": 1279,
+    "狩": 1130,
+    "独": 695,
+    "狭": 989,
+    "狼": 1575,
+    "猛": 1259,
+    "猟": 1353,
+    "猫": 1085,
+    "献": 1776,
+    "猿": 1496,
+    "獄": 1395,
+    "獣": 1958,
+    "獲": 782,
+    "玄": 1928,
+    "率": 758,
+    "玉": 1336,
+    "王": 391,
+    "珍": 1070,
+    "現": 167,
+    "球": 304,
+    "理": 116,
+    "璧": 1842,
+    "環": 659,
+    "瓦": 1995,
+    "瓶": 1387,
+    "甘": 1843,
+    "甚": 1729,
+    "生": 77,
+    "産": 406,
+    "甥": 1940,
+    "用": 138,
+    "田": 1177,
+    "由": 482,
+    "甲": 1808,
+    "申": 612,
+    "男": 759,
+    "町": 791,
+    "画": 335,
+    "界": 204,
+    "畑": 1943,
+    "留": 685,
+    "畜": 2128,
+    "略": 1125,
+    "番": 256,
+    "異": 301,
+    "疎": 1699,
+    "疑": 692,
+    "疫": 1460,
+    "疲": 1215,
+    "疼": 1619,
+    "疽": 1511,
+    "疾": 1730,
+    "病": 426,
+    "症": 549,
+    "痕": 2175,
+    "痛": 1060,
+    "痢": 1591,
+    "痩": 2151,
+    "瘍": 1529,
+    "療": 708,
+    "癌": 1428,
+    "癒": 1884,
+    "発": 83,
+    "登": 492,
+    "白": 1194,
+    "百": 1107,
+    "的": 51,
+    "皆": 1216,
+    "皇": 1535,
+    "皮": 1742,
+    "皿": 2023,
+    "盆": 2002,
+    "益": 770,
+    "盗": 1106,
+    "盛": 967,
+    "盟": 794,
+    "監": 722,
+    "盤": 1375,
+    "目": 178,
+    "盲": 1689,
+    "直": 349,
+    "相": 553,
+    "盾": 1262,
+    "省": 814,
+    "眉": 2027,
+    "看": 1407,
+    "県": 1265,
+    "真": 513,
+    "眠": 932,
+    "眺": 1068,
+    "眼": 1154,
+    "着": 448,
+    "睡": 931,
+    "督": 1348,
+    "瞑": 2266,
+    "瞬": 1745,
+    "瞭": 2019,
+    "矛": 1261,
+    "矢": 1608,
+    "知": 296,
+    "短": 1046,
+    "矯": 2080,
+    "石": 589,
+    "砂": 1108,
+    "研": 590,
+    "砕": 1292,
+    "砦": 1899,
+    "砲": 1657,
+    "破": 577,
+    "硝": 2176,
+    "硫": 1652,
+    "硬": 1245,
+    "碁": 2190,
+    "碑": 1812,
+    "確": 281,
+    "磁": 1356,
+    "磨": 2180,
+    "礁": 1271,
+    "礎": 1663,
+    "礫": 1957,
+    "示": 282,
+    "礼": 1228,
+    "社": 314,
+    "祈": 1844,
+    "祉": 1360,
+    "祖": 1163,
+    "祝": 916,
+    "神": 330,
+    "祠": 2104,
+    "祥": 1674,
+    "票": 698,
+    "祭": 944,
+    "禁": 1042,
+    "福": 1200,
+    "禽": 1390,
+    "秀": 1181,
+    "私": 361,
+    "秋": 2043,
+    "科": 429,
+    "秒": 1288,
+    "秘": 1233,
+    "秦": 2012,
+    "秩": 1184,
+    "称": 768,
+    "移": 433,
+    "稀": 1354,
+    "程": 819,
+    "税": 442,
+    "種": 319,
+    "稼": 1667,
+    "稿": 1821,
+    "穀": 1998,
+    "積": 704,
+    "穏": 1752,
+    "穫": 2039,
+    "穴": 1127,
+    "究": 628,
+    "空": 201,
+    "突": 709,
+    "窃": 1567,
+    "窒": 1313,
+    "窓": 622,
+    "窟": 911,
+    "窮": 2004,
+    "窯": 1976,
+    "立": 170,
+    "竜": 935,
+    "章": 1602,
+    "童": 1566,
+    "端": 684,
+    "競": 764,
+    "竹": 2130,
+    "笑": 1778,
+    "笛": 1875,
+    "符": 1112,
+    "第": 356,
+    "筆": 1219,
+    "等": 1111,
+    "筋": 1135,
+    "筒": 1365,
+    "答": 1007,
+    "策": 1110,
+    "箋": 1748,
+    "箔": 1990,
+    "算": 1182,
+    "管": 585,
+    "箱": 1475,
+    "節": 1556,
+    "範": 773,
+    "築": 1015,
+    "篤": 2148,
+    "簡": 641,
+    "簿": 1204,
+    "籍": 965,
+    "米": 321,
+    "粉": 1828,
+    "粋": 1863,
+    "粒": 874,
+    "粗": 2243,
+    "粘": 1813,
+    "粧": 1916,
+    "精": 809,
+    "糖": 1165,
+    "糧": 1651,
+    "糸": 1906,
+    "系": 727,
+    "糾": 2215,
+    "紀": 403,
+    "約": 221,
+    "紅": 1524,
+    "紋": 1947,
+    "納": 1277,
+    "純": 1155,
+    "紙": 895,
+    "級": 637,
+    "紛": 920,
+    "素": 452,
+    "索": 829,
+    "紫": 1903,
+    "細": 514,
+    "紹": 1857,
+    "終": 470,
+    "組": 191,
+    "絆": 2249,
+    "経": 199,
+    "結": 392,
+    "絞": 1298,
+    "絡": 1176,
+    "給": 872,
+    "絨": 2172,
+    "統": 422,
+    "絵": 1597,
+    "絶": 955,
+    "継": 646,
+    "続": 324,
+    "綜": 2251,
+    "維": 846,
+    "網": 1119,
+    "綴": 1384,
+    "綺": 2005,
+    "綻": 1482,
+    "綿": 1854,
+    "緊": 763,
+    "総": 767,
+    "緑": 1772,
+    "緒": 1605,
+    "線": 660,
+    "締": 1820,
+    "編": 1079,
+    "緩": 1334,
+    "緯": 1497,
+    "練": 1120,
+    "縁": 1794,
+    "縄": 1622,
+    "縛": 1516,
+    "縞": 1972,
+    "縦": 1211,
+    "縫": 1814,
+    "縮": 1465,
+    "績": 1026,
+    "繁": 1199,
+    "繊": 1870,
+    "繋": 1869,
+    "織": 664,
+    "繕": 2042,
+    "繞": 2116,
+    "繫": 2142,
+    "繰": 925,
+    "缶": 2212,
+    "罠": 1638,
+    "罪": 745,
+    "置": 500,
+    "罰": 1484,
+    "署": 881,
+    "罹": 2121,
+    "羅": 2179,
+    "羊": 1788,
+    "美": 924,
+    "群": 702,
+    "義": 593,
+    "羽": 1505,
+    "翌": 1625,
+    "習": 379,
+    "翻": 891,
+    "翼": 1811,
+    "老": 1793,
+    "考": 298,
+    "者": 75,
+    "耐": 1273,
+    "耕": 1512,
+    "耗": 2123,
+    "耳": 1247,
+    "聖": 978,
+    "聞": 781,
+    "聡": 1722,
+    "聴": 1310,
+    "職": 613,
+    "肉": 1461,
+    "肌": 1846,
+    "肖": 1874,
+    "肘": 2235,
+    "肝": 1104,
+    "股": 2225,
+    "肢": 807,
+    "肥": 1598,
+    "肩": 993,
+    "肪": 2199,
+    "肯": 1944,
+    "育": 504,
+    "肺": 1571,
+    "胃": 2024,
+    "胆": 1427,
+    "背": 1149,
+    "胎": 1570,
+    "胞": 756,
+    "胡": 1350,
+    "胱": 2164,
+    "胴": 2119,
+    "胸": 1969,
+    "能": 157,
+    "脂": 2017,
+    "脅": 797,
+    "脇": 1867,
+    "脈": 1072,
+    "脊": 2082,
+    "脚": 1779,
+    "脱": 1081,
+    "脳": 1047,
+    "腎": 2048,
+    "腐": 1908,
+    "腕": 1238,
+    "腫": 1332,
+    "腰": 1447,
+    "腸": 2055,
+    "腹": 1242,
+    "腺": 1945,
+    "膀": 2163,
+    "膏": 2223,
+    "膚": 1743,
+    "膜": 1600,
+    "膝": 1697,
+    "膣": 2242,
+    "膨": 1195,
+    "膳": 2194,
+    "膾": 2137,
+    "膿": 2145,
+    "臆": 1595,
+    "臓": 961,
+    "臣": 1528,
+    "臨": 1960,
+    "自": 127,
+    "臭": 2228,
+    "至": 1062,
+    "致": 1159,
+    "臼": 2054,
+    "興": 918,
+    "舌": 1926,
+    "舎": 1246,
+    "舗": 1543,
+    "舞": 921,
+    "舟": 2198,
+    "航": 284,
+    "般": 472,
+    "舶": 1735,
+    "船": 370,
+    "艇": 1462,
+    "艦": 598,
+    "良": 548,
+    "色": 1005,
+    "芝": 1302,
+    "花": 956,
+    "芳": 1950,
+    "芸": 841,
+    "芽": 1872,
+    "苔": 2088,
+    "苗": 1758,
+    "苛": 1768,
+    "若": 1092,
+    "苦": 749,
+    "英": 453,
+    "茂": 1974,
+    "茅": 2109,
+    "茎": 1946,
+    "茶": 1525,
+    "茹": 2203,
+    "草": 1250,
+    "荊": 2073,
+    "荒": 1008,
+    "荷": 750,
+    "菌": 824,
+    "菓": 1463,
+    "菜": 1837,
+    "華": 927,
+    "落": 454,
+    "葉": 603,
+    "著": 938,
+    "葛": 2184,
+    "董": 2195,
+    "葦": 2065,
+    "葬": 1102,
+    "葺": 2110,
+    "蒙": 1604,
+    "蒸": 1160,
+    "蒼": 2085,
+    "蓄": 1864,
+    "蓋": 1641,
+    "蔓": 1692,
+    "蔗": 2084,
+    "蔦": 2108,
+    "蔵": 1091,
+    "薄": 1293,
+    "薦": 1751,
+    "薬": 1033,
+    "藍": 2197,
+    "藤": 1183,
+    "藻": 1317,
+    "蘇": 2126,
+    "虎": 2144,
+    "虐": 1263,
+    "虚": 2244,
+    "虜": 1740,
+    "虫": 1239,
+    "虹": 1983,
+    "蚊": 1179,
+    "蛇": 1157,
+    "蛋": 2063,
+    "蛙": 2240,
+    "蛮": 2066,
+    "蛾": 2201,
+    "蜂": 1980,
+    "蜃": 1764,
+    "蜜": 2105,
+    "蝋": 2106,
+    "蝶": 2227,
+    "融": 607,
+    "蠣": 2057,
+    "血": 854,
+    "衆": 647,
+    "行": 71,
+    "術": 544,
+    "街": 737,
+    "衛": 736,
+    "衝": 1144,
+    "衣": 1557,
+    "表": 278,
+    "衰": 1900,
+    "袋": 1746,
+    "被": 563,
+    "裁": 582,
+    "裂": 966,
+    "装": 1174,
+    "裏": 1372,
+    "裕": 1099,
+    "補": 697,
+    "裸": 1737,
+    "製": 679,
+    "裾": 2034,
+    "複": 707,
+    "褐": 1888,
+    "褒": 1335,
+    "襲": 1141,
+    "西": 478,
+    "要": 85,
+    "覆": 757,
+    "覇": 2079,
+    "見": 123,
+    "規": 456,
+    "視": 526,
+    "覚": 605,
+    "覧": 1164,
+    "親": 530,
+    "観": 217,
+    "角": 1013,
+    "解": 292,
+    "触": 858,
+    "言": 187,
+    "訂": 1893,
+    "計": 351,
+    "討": 1319,
+    "訓": 1340,
+    "記": 297,
+    "訛": 1472,
+    "訟": 1741,
+    "訣": 1693,
+    "訪": 493,
+    "設": 262,
+    "許": 632,
+    "訳": 786,
+    "訴": 691,
+    "診": 1229,
+    "証": 459,
+    "詐": 1616,
+    "評": 772,
+    "詞": 1492,
+    "試": 376,
+    "詩": 1558,
+    "詰": 1816,
+    "話": 263,
+    "該": 988,
+    "詳": 1065,
+    "誇": 1569,
+    "誉": 1717,
+    "誌": 1138,
+    "認": 287,
+    "誓": 1934,
+    "誕": 1142,
+    "誘": 1254,
+    "語": 108,
+    "誠": 1923,
+    "誤": 1579,
+    "説": 836,
+    "読": 740,
+    "誰": 879,
+    "課": 996,
+    "調": 267,
+    "談": 1342,
+    "請": 800,
+    "論": 662,
+    "諦": 1762,
+    "諸": 792,
+    "謀": 1830,
+    "謎": 1962,
+    "謖": 2089,
+    "謙": 2028,
+    "講": 1413,
+    "謝": 844,
+    "識": 571,
+    "警": 419,
+    "議": 306,
+    "譲": 1221,
+    "護": 941,
+    "讃": 2177,
+    "讐": 1590,
+    "谷": 1124,
+    "豆": 1838,
+    "豊": 1058,
+    "豚": 1675,
+    "象": 536,
+    "豪": 1230,
+    "貌": 2188,
+    "貞": 2200,
+    "負": 491,
+    "財": 1223,
+    "貢": 1784,
+    "貧": 1326,
+    "貨": 981,
+    "販": 1386,
+    "貪": 2237,
+    "貫": 1166,
+    "責": 897,
+    "貯": 1438,
+    "貴": 1218,
+    "買": 893,
+    "貸": 1982,
+    "費": 823,
+    "貼": 1643,
+    "貿": 1097,
+    "賀": 952,
+    "賂": 1939,
+    "賃": 1048,
+    "賄": 1627,
+    "資": 441,
+    "賊": 1415,
+    "賑": 1677,
+    "賛": 1684,
+    "賞": 681,
+    "賢": 1852,
+    "質": 382,
+    "購": 640,
+    "贄": 1500,
+    "贅": 1459,
+    "贈": 1585,
+    "贖": 2239,
+    "赤": 1002,
+    "赦": 2013,
+    "走": 959,
+    "起": 387,
+    "超": 904,
+    "越": 1006,
+    "趣": 1696,
+    "足": 599,
+    "距": 696,
+    "跡": 818,
+    "跨": 2262,
+    "路": 564,
+    "跳": 1848,
+    "践": 1316,
+    "踊": 1880,
+    "踏": 1378,
+    "踵": 1212,
+    "蹄": 2131,
+    "蹴": 1942,
+    "躁": 2149,
+    "躍": 998,
+    "身": 451,
+    "車": 244,
+    "軋": 2114,
+    "軌": 1824,
+    "軍": 272,
+    "軒": 1325,
+    "軟": 1891,
+    "転": 497,
+    "軸": 1889,
+    "軻": 2074,
+    "軽": 653,
+    "較": 821,
+    "載": 1017,
+    "輝": 1859,
+    "輩": 1443,
+    "輪": 1248,
+    "輸": 728,
+    "轄": 1951,
+    "辛": 1584,
+    "辞": 1080,
+    "辣": 2222,
+    "辱": 2171,
+    "農": 699,
+    "辺": 936,
+    "込": 401,
+    "迂": 1723,
+    "迅": 1162,
+    "迎": 1695,
+    "近": 235,
+    "返": 1150,
+    "迫": 1573,
+    "述": 347,
+    "迷": 1636,
+    "追": 689,
+    "退": 1122,
+    "送": 350,
+    "逃": 855,
+    "逆": 1435,
+    "透": 1337,
+    "逐": 2226,
+    "途": 725,
+    "這": 2081,
+    "通": 119,
+    "速": 294,
+    "造": 569,
+    "連": 266,
+    "逮": 851,
+    "週": 633,
+    "進": 383,
+    "逸": 1252,
+    "遂": 1214,
+    "遅": 1053,
+    "遇": 1536,
+    "遊": 982,
+    "運": 411,
+    "過": 279,
+    "道": 258,
+    "達": 733,
+    "違": 377,
+    "遠": 732,
+    "遡": 1363,
+    "遣": 1760,
+    "遥": 2007,
+    "適": 588,
+    "遭": 1186,
+    "遮": 1653,
+    "遷": 2113,
+    "選": 196,
+    "遺": 623,
+    "遼": 2060,
+    "避": 669,
+    "還": 1063,
+    "邦": 889,
+    "邪": 1804,
+    "邸": 1382,
+    "郊": 1389,
+    "郎": 1981,
+    "郡": 1769,
+    "部": 102,
+    "郵": 1049,
+    "郷": 1473,
+    "都": 414,
+    "鄧": 1578,
+    "酋": 2238,
+    "配": 506,
+    "酒": 754,
+    "酔": 1530,
+    "酢": 2077,
+    "酬": 2033,
+    "酵": 2160,
+    "酷": 1747,
+    "酸": 985,
+    "醇": 2051,
+    "醜": 1978,
+    "醸": 1156,
+    "采": 2196,
+    "釈": 1289,
+    "里": 1234,
+    "重": 203,
+    "野": 372,
+    "量": 435,
+    "金": 206,
+    "釘": 1932,
+    "針": 964,
+    "釣": 2001,
+    "鈍": 1531,
+    "鉄": 1282,
+    "鉛": 1429,
+    "鉱": 1789,
+    "銀": 667,
+    "銃": 1311,
+    "銅": 1637,
+    "銘": 2192,
+    "鋤": 2068,
+    "鋭": 1400,
+    "鋼": 1468,
+    "錆": 1712,
+    "錠": 1910,
+    "錦": 1351,
+    "錨": 2246,
+    "錬": 1858,
+    "錯": 1422,
+    "録": 380,
+    "鍛": 1498,
+    "鍵": 1208,
+    "鎖": 1034,
+    "鎧": 1609,
+    "鎮": 2186,
+    "鏡": 834,
+    "鐘": 1970,
+    "鐙": 1540,
+    "鑑": 1515,
+    "長": 177,
+    "門": 799,
+    "閉": 780,
+    "開": 195,
+    "閑": 2117,
+    "間": 97,
+    "関": 202,
+    "閣": 1952,
+    "閥": 2247,
+    "閲": 1300,
+    "闇": 1898,
+    "闘": 1180,
+    "阪": 1410,
+    "防": 557,
+    "阻": 1207,
+    "附": 1953,
+    "降": 484,
+    "限": 385,
+    "陛": 2062,
+    "院": 374,
+    "陣": 1145,
+    "除": 700,
+    "陥": 971,
+    "陰": 1831,
+    "陳": 1894,
+    "陵": 1727,
+    "陶": 1504,
+    "陸": 410,
+    "険": 436,
+    "陽": 591,
+    "隅": 1971,
+    "隊": 357,
+    "階": 769,
+    "随": 1989,
+    "隔": 1269,
+    "隕": 1434,
+    "隙": 1403,
+    "際": 295,
+    "障": 687,
+    "隠": 1161,
+    "隣": 934,
+    "隷": 1826,
+    "隻": 1076,
+    "雀": 1552,
+    "雄": 1718,
+    "雅": 1620,
+    "集": 253,
+    "雇": 1191,
+    "雌": 2191,
+    "雑": 686,
+    "離": 412,
+    "難": 558,
+    "雨": 499,
+    "雪": 665,
+    "雰": 1572,
+    "雲": 1178,
+    "零": 2165,
+    "雷": 1198,
+    "電": 252,
+    "需": 1576,
+    "震": 739,
+    "霊": 1306,
+    "霧": 1709,
+    "露": 825,
+    "青": 1369,
+    "静": 1417,
+    "非": 243,
+    "面": 255,
+    "革": 850,
+    "靭": 1406,
+    "靴": 1116,
+    "鞍": 1541,
+    "鞭": 2166,
+    "韓": 1299,
+    "音": 509,
+    "韻": 2159,
+    "響": 432,
+    "頂": 1227,
+    "頃": 560,
+    "項": 1862,
+    "順": 1023,
+    "須": 1430,
+    "預": 1542,
+    "頑": 1877,
+    "領": 458,
+    "頬": 1959,
+    "頭": 734,
+    "頸": 1913,
+    "頻": 990,
+    "頼": 1020,
+    "顆": 2008,
+    "題": 384,
+    "額": 980,
+    "顎": 1534,
+    "顔": 1222,
+    "顕": 1346,
+    "願": 1593,
+    "類": 308,
+    "顧": 1093,
+    "風": 490,
+    "飛": 592,
+    "食": 283,
+    "飢": 1876,
+    "飯": 1841,
+    "飲": 611,
+    "飼": 1290,
+    "飽": 1840,
+    "飾": 1043,
+    "養": 1366,
+    "餌": 1494,
+    "餐": 2183,
+    "餓": 2155,
+    "館": 885,
+    "饉": 2214,
+    "首": 529,
+    "香": 1158,
+    "馬": 638,
+    "馳": 1560,
+    "馴": 1188,
+    "駄": 1817,
+    "駅": 1771,
+    "駆": 1132,
+    "駈": 2250,
+    "駐": 1143,
+    "騎": 1121,
+    "騒": 1822,
+    "験": 345,
+    "騙": 1703,
+    "騰": 2207,
+    "驚": 1220,
+    "骨": 1100,
+    "骸": 1645,
+    "高": 173,
+    "髪": 1782,
+    "髭": 2101,
+    "鬱": 2150,
+    "魂": 1975,
+    "魅": 954,
+    "魔": 1786,
+    "魚": 1367,
+    "鮫": 1283,
+    "鮮": 1358,
+    "鳥": 626,
+    "鳩": 2129,
+    "鳴": 1355,
+    "鴨": 2218,
+    "鶏": 1845,
+    "鶯": 2086,
+    "鷹": 2173,
+    "鹿": 1470,
+    "麓": 1331,
+    "麗": 2006,
+    "麦": 1798,
+    "麻": 1359,
+    "黄": 1392,
+    "黒": 1040,
+    "黙": 1690,
+    "鼓": 1905,
+    "鼻": 1610,
+    "齢": 1059,
+    "齧": 1553,
+    "龍": 2158,
+    "！": 1445,
+    "１": 1728,
+    "～": 501
+  },
+  "jun": {
+    "'": 37,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଁ": 36,
+    "ଂ": 38,
+    "ଅ": 26,
+    "ଆ": 11,
+    "ଇ": 24,
+    "ଉ": 28,
+    "ଏ": 21,
+    "ଓ": 46,
+    "କ": 8,
+    "ଖ": 47,
+    "ଗ": 19,
+    "ଙ": 27,
+    "ଚ": 32,
+    "ଜ": 18,
+    "ଞ": 22,
+    "ଟ": 33,
+    "ଡ": 23,
+    "ଣ": 31,
+    "ତ": 14,
+    "ଥ": 56,
+    "ଦ": 30,
+    "ଧ": 42,
+    "ନ": 13,
+    "ପ": 25,
+    "ଫ": 55,
+    "ବ": 16,
+    "ମ": 17,
+    "ର": 9,
+    "ଲ": 20,
+    "ଳ": 29,
+    "ଶ": 44,
+    "ଷ": 45,
+    "ସ": 15,
+    "ହ": 39,
+    "଼": 50,
+    "ା": 6,
+    "ି": 5,
+    "ୁ": 12,
+    "େ": 7,
+    "୍": 10,
+    "ୟ": 34,
+    "୦": 41,
+    "୧": 48,
+    "୨": 54,
+    "୩": 52,
+    "୪": 43,
+    "୫": 49,
+    "୬": 51,
+    "୯": 53,
+    "‍": 35
+  },
+  "juy": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ଁ": 37,
+    "ଂ": 39,
+    "ଃ": 33,
+    "ଅ": 24,
+    "ଆ": 9,
+    "ଇ": 30,
+    "ଉ": 36,
+    "ଏ": 28,
+    "ଓ": 35,
+    "କ": 25,
+    "ଗ": 26,
+    "ଙ": 14,
+    "ଜ": 13,
+    "ଞ": 20,
+    "ଟ": 38,
+    "ଡ": 10,
+    "ଣ": 32,
+    "ତ": 15,
+    "ଦ": 34,
+    "ନ": 6,
+    "ପ": 27,
+    "ବ": 18,
+    "ମ": 16,
+    "ର": 12,
+    "ଲ": 17,
+    "ସ": 19,
+    "଼": 29,
+    "ା": 11,
+    "ି": 7,
+    "ୁ": 23,
+    "େ": 8,
+    "ୋ": 22,
+    "୍": 4,
+    "ୟ": 21,
+    "ୱ": 31,
+    "‍": 41
+  },
+  "jvn": {
+    "'": 31,
+    "-": 27,
+    "0": 30,
+    "1": 34,
+    "2": 35,
+    "5": 36,
+    "6": 37,
+    "8": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "d": 19,
+    "e": 17,
+    "f": 28,
+    "g": 7,
+    "h": 25,
+    "i": 9,
+    "j": 26,
+    "k": 8,
+    "l": 16,
+    "m": 20,
+    "n": 6,
+    "o": 15,
+    "p": 21,
+    "r": 13,
+    "s": 11,
+    "t": 14,
+    "u": 10,
+    "v": 29,
+    "w": 18,
+    "y": 24,
+    "z": 32,
+    "|": 4,
+    "è": 23,
+    "é": 12
+  },
+  "kaa": {
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "μ": 36,
+    "а": 5,
+    "б": 16,
+    "в": 43,
+    "г": 30,
+    "д": 12,
+    "е": 6,
+    "ж": 28,
+    "з": 22,
+    "и": 7,
+    "й": 17,
+    "к": 23,
+    "л": 10,
+    "м": 15,
+    "н": 9,
+    "о": 18,
+    "п": 20,
+    "р": 11,
+    "с": 13,
+    "т": 14,
+    "у": 21,
+    "ф": 46,
+    "х": 34,
+    "ц": 49,
+    "ш": 24,
+    "ы": 8,
+    "ь": 45,
+    "э": 48,
+    "ю": 47,
+    "я": 38,
+    "ѓ": 27,
+    "є": 29,
+    "њ": 25,
+    "ќ": 19,
+    "ў": 26,
+    "ѳ": 40,
+    "ғ": 37,
+    "қ": 33,
+    "ң": 35,
+    "ү": 32,
+    "ҳ": 31,
+    "ә": 41,
+    "ө": 42,
+    "–": 39
+  },
+  "kab": {
+    "'": 46,
+    "-": 16,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 27,
+    "d": 10,
+    "e": 6,
+    "f": 22,
+    "g": 23,
+    "h": 33,
+    "i": 7,
+    "j": 36,
+    "k": 19,
+    "l": 12,
+    "m": 13,
+    "n": 9,
+    "o": 39,
+    "p": 41,
+    "q": 24,
+    "r": 14,
+    "s": 11,
+    "t": 8,
+    "u": 15,
+    "v": 42,
+    "w": 20,
+    "x": 30,
+    "y": 17,
+    "z": 26,
+    "|": 4,
+    "â": 60,
+    "æ": 64,
+    "è": 52,
+    "é": 48,
+    "î": 57,
+    "ï": 47,
+    "ó": 56,
+    "ĉ": 49,
+    "č": 35,
+    "ĝ": 50,
+    "ğ": 44,
+    "ĥ": 61,
+    "ţ": 51,
+    "ǧ": 37,
+    "ɛ": 29,
+    "ɣ": 18,
+    "̣": 63,
+    "γ": 43,
+    "ε": 40,
+    "σ": 59,
+    "ԑ": 45,
+    "ḅ": 53,
+    "ḍ": 25,
+    "ḏ": 58,
+    "ḥ": 28,
+    "ṉ": 62,
+    "ṛ": 31,
+    "ṣ": 38,
+    "ṭ": 32,
+    "ẓ": 34,
+    "‐": 54,
+    "–": 55
+  },
+  "kac": {
+    "'": 28,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 25,
+    "d": 18,
+    "e": 13,
+    "g": 9,
+    "h": 8,
+    "i": 7,
+    "j": 22,
+    "k": 16,
+    "l": 20,
+    "m": 10,
+    "n": 6,
+    "o": 24,
+    "p": 21,
+    "r": 17,
+    "s": 14,
+    "t": 12,
+    "u": 11,
+    "w": 15,
+    "y": 19,
+    "z": 26,
+    "|": 4
+  },
+  "kak": {
+    "'": 31,
+    "-": 20,
+    "0": 35,
+    "1": 37,
+    "2": 39,
+    "6": 33,
+    "7": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 27,
+    "d": 11,
+    "e": 16,
+    "f": 28,
+    "g": 10,
+    "h": 12,
+    "i": 7,
+    "j": 26,
+    "k": 17,
+    "l": 15,
+    "m": 14,
+    "n": 6,
+    "o": 8,
+    "p": 18,
+    "q": 36,
+    "r": 25,
+    "s": 22,
+    "t": 9,
+    "u": 24,
+    "v": 30,
+    "w": 21,
+    "x": 32,
+    "y": 13,
+    "z": 29,
+    "|": 4,
+    "â": 23,
+    "ñ": 34
+  },
+  "kam": {
+    "!": 67,
+    "$": 54,
+    "%": 55,
+    "&": 63,
+    "'": 26,
+    "+": 59,
+    ",": 47,
+    "-": 44,
+    ".": 43,
+    "/": 53,
+    "0": 30,
+    "1": 31,
+    "2": 34,
+    "3": 41,
+    "4": 39,
+    "5": 38,
+    "6": 42,
+    "7": 45,
+    "8": 40,
+    "9": 37,
+    ":": 52,
+    ";": 57,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 27,
+    "d": 21,
+    "e": 9,
+    "f": 32,
+    "g": 23,
+    "h": 19,
+    "i": 6,
+    "j": 35,
+    "k": 8,
+    "l": 17,
+    "m": 11,
+    "n": 7,
+    "o": 15,
+    "p": 28,
+    "q": 49,
+    "r": 25,
+    "s": 16,
+    "t": 12,
+    "u": 10,
+    "v": 22,
+    "w": 13,
+    "x": 46,
+    "y": 14,
+    "z": 29,
+    "|": 4,
+    "~": 58,
+    "£": 62,
+    "°": 64,
+    "á": 66,
+    "ã": 65,
+    "ì": 60,
+    "í": 48,
+    "î": 33,
+    "ú": 50,
+    "û": 36,
+    "ü": 56,
+    "ĩ": 18,
+    "ũ": 20,
+    "ḉ": 61,
+    "”": 51
+  },
+  "kan": {
+    "!": 110,
+    "\"": 91,
+    "$": 105,
+    "%": 106,
+    "'": 97,
+    "+": 117,
+    ",": 69,
+    "-": 63,
+    ".": 57,
+    "/": 85,
+    "0": 48,
+    "1": 52,
+    "2": 54,
+    "3": 65,
+    "4": 62,
+    "5": 60,
+    "6": 64,
+    "7": 67,
+    "8": 66,
+    "9": 58,
+    ":": 79,
+    ";": 104,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 115,
+    "[": 123,
+    "]": 124,
+    "a": 74,
+    "b": 95,
+    "c": 76,
+    "d": 89,
+    "e": 86,
+    "f": 107,
+    "g": 99,
+    "h": 102,
+    "i": 75,
+    "j": 120,
+    "k": 108,
+    "l": 93,
+    "m": 80,
+    "n": 87,
+    "o": 77,
+    "p": 84,
+    "q": 119,
+    "r": 82,
+    "s": 83,
+    "t": 78,
+    "u": 88,
+    "v": 96,
+    "w": 103,
+    "x": 109,
+    "y": 111,
+    "z": 113,
+    "|": 4,
+    "°": 114,
+    "²": 121,
+    "õ": 116,
+    "ಂ": 21,
+    "ಃ": 81,
+    "ಅ": 29,
+    "ಆ": 39,
+    "ಇ": 43,
+    "ಈ": 53,
+    "ಉ": 50,
+    "ಊ": 92,
+    "ಋ": 100,
+    "ಎ": 42,
+    "ಏ": 59,
+    "ಐ": 68,
+    "ಒ": 47,
+    "ಓ": 73,
+    "ಔ": 90,
+    "ಕ": 17,
+    "ಖ": 51,
+    "ಗ": 14,
+    "ಘ": 61,
+    "ಚ": 31,
+    "ಛ": 94,
+    "ಜ": 35,
+    "ಝ": 101,
+    "ಞ": 72,
+    "ಟ": 26,
+    "ಠ": 71,
+    "ಡ": 24,
+    "ಢ": 98,
+    "ಣ": 37,
+    "ತ": 12,
+    "ಥ": 45,
+    "ದ": 10,
+    "ಧ": 40,
+    "ನ": 11,
+    "ಪ": 23,
+    "ಫ": 49,
+    "ಬ": 27,
+    "ಭ": 44,
+    "ಮ": 20,
+    "ಯ": 19,
+    "ರ": 8,
+    "ಲ": 16,
+    "ಳ": 22,
+    "ವ": 15,
+    "ಶ": 38,
+    "ಷ": 36,
+    "ಸ": 18,
+    "ಹ": 25,
+    "ಾ": 9,
+    "ಿ": 6,
+    "ೀ": 32,
+    "ು": 7,
+    "ೂ": 33,
+    "ೃ": 55,
+    "ೆ": 13,
+    "ೇ": 28,
+    "ೈ": 46,
+    "ೊ": 30,
+    "ೋ": 34,
+    "ೌ": 56,
+    "್": 5,
+    "೪": 122,
+    "​": 112,
+    "‌": 41,
+    "‍": 70,
+    "–": 118
+  },
+  "kao": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 33,
+    "d": 18,
+    "e": 10,
+    "f": 21,
+    "g": 25,
+    "h": 30,
+    "i": 8,
+    "j": 27,
+    "k": 17,
+    "l": 9,
+    "m": 14,
+    "n": 6,
+    "o": 7,
+    "p": 31,
+    "r": 20,
+    "s": 16,
+    "t": 13,
+    "u": 11,
+    "w": 19,
+    "x": 12,
+    "y": 22,
+    "|": 4,
+    "­": 34,
+    "é": 32,
+    "í": 28,
+    "ń": 29,
+    "ŋ": 26,
+    "ɲ": 23
+  },
+  "kaq": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 15,
+    "d": 24,
+    "e": 10,
+    "f": 27,
+    "g": 28,
+    "h": 8,
+    "i": 6,
+    "j": 13,
+    "k": 36,
+    "l": 25,
+    "m": 18,
+    "n": 7,
+    "o": 9,
+    "p": 19,
+    "q": 17,
+    "r": 20,
+    "s": 14,
+    "t": 12,
+    "u": 11,
+    "v": 30,
+    "x": 22,
+    "y": 21,
+    "z": 31,
+    "|": 5,
+    "á": 29,
+    "é": 32,
+    "í": 26,
+    "ñ": 35,
+    "ó": 33,
+    "ú": 34,
+    "—": 23
+  },
+  "kat": {
+    "!": 90,
+    "%": 79,
+    "&": 87,
+    "'": 83,
+    "+": 84,
+    ",": 63,
+    "-": 34,
+    ".": 59,
+    "/": 65,
+    "0": 36,
+    "1": 37,
+    "2": 40,
+    "3": 48,
+    "4": 43,
+    "5": 44,
+    "6": 46,
+    "7": 56,
+    "8": 51,
+    "9": 41,
+    ":": 67,
+    ";": 77,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 88,
+    "]": 89,
+    "a": 42,
+    "b": 70,
+    "c": 55,
+    "d": 61,
+    "e": 49,
+    "f": 72,
+    "g": 62,
+    "h": 66,
+    "i": 50,
+    "j": 78,
+    "k": 71,
+    "l": 58,
+    "m": 64,
+    "n": 45,
+    "o": 47,
+    "p": 60,
+    "q": 81,
+    "r": 54,
+    "s": 52,
+    "t": 57,
+    "u": 68,
+    "v": 69,
+    "w": 74,
+    "x": 76,
+    "y": 75,
+    "z": 82,
+    "|": 4,
+    "°": 85,
+    "²": 91,
+    "õ": 80,
+    "ú": 86,
+    "ა": 5,
+    "ბ": 15,
+    "გ": 19,
+    "დ": 13,
+    "��": 7,
+    "ვ": 16,
+    "ზ": 25,
+    "თ": 18,
+    "ი": 6,
+    "კ": 24,
+    "ლ": 14,
+    "მ": 11,
+    "ნ": 12,
+    "ო": 10,
+    "პ": 28,
+    "ჟ": 53,
+    "რ": 9,
+    "ს": 8,
+    "ტ": 21,
+    "უ": 17,
+    "ფ": 29,
+    "ქ": 26,
+    "ღ": 32,
+    "ყ": 30,
+    "შ": 20,
+    "ჩ": 33,
+    "ც": 23,
+    "ძ": 31,
+    "წ": 27,
+    "ჭ": 39,
+    "ხ": 22,
+    "ჯ": 35,
+    "ჰ": 38,
+    "–": 92,
+    "—": 73
+  },
+  "kay": {
+    "'": 11,
+    "0": 35,
+    "1": 36,
+    "2": 39,
+    "3": 44,
+    "5": 40,
+    "6": 46,
+    "7": 41,
+    "8": 43,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 33,
+    "d": 29,
+    "e": 6,
+    "f": 37,
+    "g": 23,
+    "h": 21,
+    "i": 16,
+    "j": 9,
+    "k": 10,
+    "l": 28,
+    "m": 13,
+    "n": 15,
+    "o": 7,
+    "p": 18,
+    "r": 8,
+    "s": 20,
+    "t": 14,
+    "u": 19,
+    "v": 38,
+    "w": 12,
+    "x": 47,
+    "y": 17,
+    "z": 34,
+    "|": 4,
+    "ã": 25,
+    "ç": 45,
+    "õ": 27,
+    "ĩ": 30,
+    "ũ": 31,
+    "ẽ": 22,
+    "ỹ": 26,
+    "—": 24
+  },
+  "kaz": {
+    "!": 92,
+    "$": 101,
+    "%": 89,
+    "&": 93,
+    "'": 88,
+    "+": 95,
+    ",": 81,
+    "-": 37,
+    ".": 59,
+    "/": 79,
+    "0": 39,
+    "1": 41,
+    "2": 46,
+    "3": 61,
+    "4": 57,
+    "5": 56,
+    "6": 63,
+    "7": 69,
+    "8": 62,
+    "9": 55,
+    ":": 74,
+    ";": 91,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 103,
+    "[": 99,
+    "]": 100,
+    "a": 43,
+    "b": 73,
+    "c": 60,
+    "d": 68,
+    "e": 45,
+    "f": 76,
+    "g": 71,
+    "h": 66,
+    "i": 51,
+    "j": 85,
+    "k": 75,
+    "l": 58,
+    "m": 72,
+    "n": 47,
+    "o": 48,
+    "p": 67,
+    "q": 90,
+    "r": 50,
+    "s": 54,
+    "t": 49,
+    "u": 70,
+    "v": 80,
+    "w": 77,
+    "x": 84,
+    "y": 78,
+    "z": 86,
+    "|": 4,
+    "«": 52,
+    "°": 97,
+    "»": 53,
+    "í": 104,
+    "ó": 105,
+    "õ": 106,
+    "ú": 102,
+    "а": 5,
+    "б": 19,
+    "в": 35,
+    "г": 27,
+    "д": 13,
+    "е": 6,
+    "ж": 22,
+    "з": 28,
+    "и": 21,
+    "й": 24,
+    "к": 15,
+    "л": 12,
+    "м": 17,
+    "н": 7,
+    "о": 16,
+    "п": 25,
+    "р": 10,
+    "с": 14,
+    "т": 9,
+    "у": 20,
+    "ф": 36,
+    "х": 38,
+    "ц": 40,
+    "ч": 65,
+    "ш": 29,
+    "щ": 96,
+    "ъ": 83,
+    "ы": 8,
+    "ь": 42,
+    "э": 44,
+    "ю": 64,
+    "я": 34,
+    "ё": 107,
+    "і": 11,
+    "ғ": 23,
+    "қ": 18,
+    "ң": 26,
+    "ү": 30,
+    "ұ": 33,
+    "һ": 87,
+    "ә": 31,
+    "ө": 32,
+    "–": 94,
+    "—": 82,
+    "№": 98
+  },
+  "kbo": {
+    "'": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 28,
+    "b": 20,
+    "c": 43,
+    "d": 19,
+    "e": 37,
+    "f": 45,
+    "g": 18,
+    "h": 49,
+    "i": 7,
+    "j": 40,
+    "k": 13,
+    "l": 32,
+    "m": 15,
+    "n": 8,
+    "o": 39,
+    "p": 23,
+    "r": 5,
+    "s": 25,
+    "t": 26,
+    "u": 35,
+    "v": 31,
+    "w": 47,
+    "y": 27,
+    "z": 29,
+    "|": 4,
+    "á": 10,
+    "â": 44,
+    "ã": 11,
+    "é": 21,
+    "í": 9,
+    "î": 48,
+    "ó": 38,
+    "ô": 50,
+    "õ": 36,
+    "ú": 14,
+    "ĩ": 6,
+    "ŋ": 46,
+    "ũ": 30,
+    "́": 16,
+    "̂": 51,
+    "̃": 24,
+    "ẹ": 34,
+    "ẽ": 33,
+    "ị": 17,
+    "ọ": 42,
+    "ộ": 41,
+    "ụ": 22,
+    "ꞌ": 12
+  },
+  "kbp": {
+    "-": 19,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 33,
+    "d": 30,
+    "e": 12,
+    "f": 38,
+    "g": 35,
+    "h": 31,
+    "i": 20,
+    "j": 39,
+    "k": 18,
+    "l": 15,
+    "m": 14,
+    "n": 11,
+    "o": 29,
+    "p": 13,
+    "r": 36,
+    "s": 17,
+    "t": 16,
+    "u": 27,
+    "v": 42,
+    "w": 22,
+    "y": 10,
+    "z": 28,
+    "|": 4,
+    "á": 34,
+    "é": 45,
+    "ñ": 32,
+    "ó": 43,
+    "ú": 44,
+    "đ": 41,
+    "ń": 40,
+    "ŋ": 24,
+    "ɔ": 8,
+    "ɖ": 21,
+    "ɛ": 6,
+    "ɣ": 23,
+    "ɩ": 7,
+    "ʋ": 9,
+    "́": 26,
+    "ḿ": 37
+  },
+  "kbq": {
+    "'": 11,
+    "-": 29,
+    "0": 28,
+    "1": 27,
+    "2": 31,
+    "3": 34,
+    "4": 33,
+    "5": 35,
+    "6": 36,
+    "7": 32,
+    "8": 37,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "d": 25,
+    "e": 7,
+    "f": 21,
+    "g": 15,
+    "h": 13,
+    "i": 9,
+    "j": 23,
+    "k": 18,
+    "l": 24,
+    "m": 8,
+    "n": 6,
+    "o": 10,
+    "p": 22,
+    "q": 40,
+    "r": 14,
+    "s": 20,
+    "t": 17,
+    "u": 12,
+    "v": 19,
+    "w": 30,
+    "x": 41,
+    "y": 39,
+    "z": 16,
+    "|": 5
+  },
+  "kbr": {
+    "'": 24,
+    "-": 30,
+    "1": 33,
+    "4": 34,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 12,
+    "c": 13,
+    "d": 23,
+    "e": 8,
+    "f": 27,
+    "g": 19,
+    "h": 10,
+    "i": 7,
+    "j": 28,
+    "k": 20,
+    "l": 18,
+    "m": 16,
+    "n": 9,
+    "o": 5,
+    "p": 29,
+    "q": 22,
+    "r": 21,
+    "s": 15,
+    "t": 11,
+    "u": 17,
+    "w": 25,
+    "x": 26,
+    "y": 14,
+    "z": 31,
+    "|": 4
+  },
+  "kby": {
+    "'": 30,
+    "-": 26,
+    "3": 32,
+    "4": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 24,
+    "d": 9,
+    "e": 11,
+    "f": 25,
+    "g": 22,
+    "h": 27,
+    "i": 7,
+    "j": 18,
+    "k": 14,
+    "l": 19,
+    "m": 17,
+    "n": 6,
+    "o": 8,
+    "p": 28,
+    "r": 12,
+    "s": 20,
+    "t": 21,
+    "u": 10,
+    "w": 15,
+    "y": 16,
+    "|": 5,
+    "ǝ": 13,
+    "ɍ": 29
+  },
+  "kca": {
+    "'": 30,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ă": 48,
+    "ŏ": 19,
+    "а": 7,
+    "б": 41,
+    "в": 20,
+    "г": 36,
+    "д": 34,
+    "е": 25,
+    "з": 37,
+    "и": 13,
+    "й": 11,
+    "к": 26,
+    "л": 31,
+    "м": 10,
+    "н": 9,
+    "о": 28,
+    "п": 16,
+    "р": 21,
+    "с": 14,
+    "т": 5,
+    "у": 18,
+    "ф": 39,
+    "х": 17,
+    "ц": 46,
+    "ч": 45,
+    "ш": 29,
+    "щ": 22,
+    "ы": 12,
+    "ь": 43,
+    "э": 23,
+    "ю": 47,
+    "я": 38,
+    "ё": 44,
+    "ў": 24,
+    "ӈ": 27,
+    "ӑ": 15,
+    "ә": 8,
+    "ԉ": 33,
+    "ԋ": 32,
+    "ԏ": 42,
+    "ԓ": 6,
+    "–": 35
+  },
+  "kcg": {
+    "'": 34,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 28,
+    "d": 22,
+    "e": 20,
+    "f": 26,
+    "g": 9,
+    "h": 17,
+    "i": 8,
+    "j": 25,
+    "k": 13,
+    "l": 24,
+    "m": 15,
+    "n": 7,
+    "o": 19,
+    "p": 27,
+    "r": 29,
+    "s": 16,
+    "t": 11,
+    "u": 12,
+    "v": 30,
+    "w": 14,
+    "y": 10,
+    "z": 21,
+    "|": 4,
+    "á": 23,
+    "é": 33,
+    "í": 32,
+    "ó": 31,
+    "ú": 35,
+    "̱": 6
+  },
+  "kdc": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 27,
+    "d": 19,
+    "e": 10,
+    "f": 28,
+    "g": 15,
+    "h": 16,
+    "i": 6,
+    "j": 24,
+    "k": 13,
+    "l": 9,
+    "m": 14,
+    "n": 8,
+    "o": 11,
+    "p": 26,
+    "s": 18,
+    "t": 21,
+    "u": 7,
+    "v": 23,
+    "w": 12,
+    "y": 17,
+    "z": 22,
+    "|": 5
+  },
+  "kde": {
+    "'": 27,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "d": 16,
+    "e": 12,
+    "g": 17,
+    "h": 21,
+    "i": 8,
+    "j": 24,
+    "k": 10,
+    "l": 9,
+    "m": 11,
+    "n": 6,
+    "o": 13,
+    "p": 22,
+    "s": 20,
+    "t": 18,
+    "u": 7,
+    "v": 15,
+    "w": 14,
+    "y": 19,
+    "|": 5,
+    "à": 31,
+    "á": 25,
+    "è": 35,
+    "é": 28,
+    "ì": 34,
+    "í": 26,
+    "ó": 32,
+    "ù": 33,
+    "ú": 29
+  },
+  "kdh": {
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 12,
+    "c": 40,
+    "d": 16,
+    "e": 21,
+    "f": 37,
+    "g": 25,
+    "h": 43,
+    "i": 27,
+    "j": 38,
+    "k": 19,
+    "l": 17,
+    "m": 14,
+    "n": 9,
+    "o": 29,
+    "p": 42,
+    "r": 20,
+    "s": 13,
+    "t": 23,
+    "u": 33,
+    "v": 35,
+    "w": 15,
+    "y": 24,
+    "z": 28,
+    "|": 4,
+    "á": 10,
+    "é": 31,
+    "í": 26,
+    "ó": 39,
+    "ú": 32,
+    "ÿ": 45,
+    "ń": 36,
+    "ŋ": 34,
+    "ɔ": 11,
+    "ɖ": 22,
+    "ɛ": 8,
+    "ɩ": 5,
+    "ʊ": 18,
+    "́": 6,
+    "ḿ": 41,
+    "‐": 44
+  },
+  "kdi": {
+    "'": 25,
+    "-": 29,
+    "0": 28,
+    "3": 27,
+    "4": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 17,
+    "c": 22,
+    "d": 12,
+    "e": 6,
+    "g": 15,
+    "i": 8,
+    "j": 24,
+    "k": 9,
+    "l": 19,
+    "m": 11,
+    "n": 10,
+    "o": 5,
+    "p": 21,
+    "r": 16,
+    "s": 23,
+    "t": 14,
+    "u": 13,
+    "w": 18,
+    "y": 20,
+    "|": 4
+  },
+  "kdj": {
+    "'": 28,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 23,
+    "d": 20,
+    "e": 7,
+    "g": 25,
+    "i": 6,
+    "j": 24,
+    "k": 9,
+    "l": 15,
+    "m": 18,
+    "n": 10,
+    "o": 8,
+    "p": 19,
+    "r": 14,
+    "s": 17,
+    "t": 13,
+    "u": 12,
+    "w": 22,
+    "y": 16,
+    "|": 4,
+    "ŋ": 11,
+    "–": 26
+  },
+  "kdl": {
+    "'": 35,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 26,
+    "d": 27,
+    "e": 13,
+    "f": 31,
+    "g": 19,
+    "h": 21,
+    "i": 7,
+    "j": 32,
+    "k": 9,
+    "l": 17,
+    "m": 14,
+    "n": 8,
+    "o": 16,
+    "p": 24,
+    "r": 20,
+    "s": 11,
+    "t": 12,
+    "u": 6,
+    "v": 25,
+    "w": 18,
+    "y": 15,
+    "z": 28,
+    "|": 4,
+    "ɓ": 30,
+    "ɗ": 22,
+    "̱": 10,
+    "‐": 33,
+    "ꞌ": 29
+  },
+  "kdn": {
+    "'": 30,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 25,
+    "d": 13,
+    "e": 10,
+    "f": 26,
+    "g": 23,
+    "h": 17,
+    "i": 6,
+    "j": 27,
+    "k": 9,
+    "l": 12,
+    "m": 11,
+    "n": 8,
+    "o": 14,
+    "p": 20,
+    "r": 28,
+    "s": 21,
+    "t": 16,
+    "u": 7,
+    "v": 24,
+    "w": 15,
+    "y": 19,
+    "z": 18,
+    "|": 5,
+    "–": 31
+  },
+  "kdt": {
+    "-": 62,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 8,
+    "ក": 4,
+    "ខ": 48,
+    "គ": 44,
+    "ឃ": 50,
+    "ង": 9,
+    "ច": 19,
+    "ឆ": 53,
+    "ជ": 39,
+    "ឈ": 59,
+    "ញ": 46,
+    "ដ": 38,
+    "ឋ": 60,
+    "ឌ": 54,
+    "ណ": 26,
+    "ត": 16,
+    "ថ": 51,
+    "ទ": 33,
+    "ធ": 55,
+    "ន": 15,
+    "ប": 12,
+    "ផ": 49,
+    "ព": 22,
+    "ភ": 57,
+    "ម": 14,
+    "យ": 13,
+    "រ": 7,
+    "ល": 27,
+    "វ": 18,
+    "ស": 11,
+    "ហ": 41,
+    "ឡ": 45,
+    "អ": 17,
+    "ឥ": 61,
+    "ឱ": 63,
+    "ា": 5,
+    "ិ": 28,
+    "ី": 42,
+    "ឹ": 20,
+    "ឺ": 24,
+    "ុ": 43,
+    "ូ": 29,
+    "ួ": 30,
+    "ើ": 21,
+    "ឿ": 58,
+    "ៀ": 40,
+    "េ": 23,
+    "ែ": 32,
+    "ៃ": 36,
+    "ោ": 35,
+    "ៅ": 56,
+    "ំ": 37,
+    "ះ": 31,
+    "៉": 10,
+    "៊": 47,
+    "់": 25,
+    "៍": 52,
+    "័": 34,
+    "្": 6
+  },
+  "kea": {
+    "$": 66,
+    "%": 58,
+    "&": 68,
+    "'": 54,
+    "+": 62,
+    ",": 52,
+    "-": 36,
+    ".": 44,
+    "/": 51,
+    "0": 32,
+    "1": 37,
+    "2": 39,
+    "3": 45,
+    "4": 43,
+    "5": 42,
+    "6": 46,
+    "7": 48,
+    "8": 47,
+    "9": 41,
+    ":": 50,
+    ";": 60,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 63,
+    "]": 64,
+    "a": 5,
+    "b": 19,
+    "c": 35,
+    "d": 12,
+    "e": 11,
+    "f": 20,
+    "g": 22,
+    "h": 27,
+    "i": 6,
+    "j": 25,
+    "k": 15,
+    "l": 16,
+    "m": 18,
+    "n": 7,
+    "o": 14,
+    "p": 17,
+    "q": 53,
+    "r": 13,
+    "s": 9,
+    "t": 10,
+    "u": 8,
+    "v": 21,
+    "w": 40,
+    "x": 29,
+    "y": 26,
+    "z": 24,
+    "|": 4,
+    "£": 72,
+    "¥": 65,
+    "ª": 67,
+    "°": 69,
+    "²": 73,
+    "º": 56,
+    "á": 28,
+    "â": 49,
+    "ã": 59,
+    "ç": 71,
+    "é": 23,
+    "ê": 34,
+    "í": 30,
+    "ó": 33,
+    "ô": 31,
+    "õ": 61,
+    "ú": 38,
+    "ü": 57,
+    "ł": 70,
+    "̇": 74,
+    "”": 55
+  },
+  "kek": {
+    "'": 11,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 8,
+    "d": 26,
+    "e": 12,
+    "f": 36,
+    "g": 33,
+    "h": 18,
+    "i": 6,
+    "j": 21,
+    "k": 19,
+    "l": 9,
+    "m": 24,
+    "n": 10,
+    "o": 17,
+    "p": 27,
+    "q": 23,
+    "r": 20,
+    "s": 22,
+    "t": 13,
+    "u": 7,
+    "v": 38,
+    "x": 16,
+    "y": 25,
+    "z": 28,
+    "|": 4,
+    "á": 37,
+    "é": 35,
+    "í": 32,
+    "ñ": 39,
+    "ó": 34,
+    "ú": 31,
+    "̱": 14,
+    "—": 29
+  },
+  "ken": {
+    "'": 50,
+    "-": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 7,
+    "c": 23,
+    "d": 24,
+    "e": 32,
+    "f": 33,
+    "g": 31,
+    "h": 13,
+    "i": 18,
+    "j": 42,
+    "k": 10,
+    "l": 39,
+    "m": 9,
+    "n": 6,
+    "o": 25,
+    "p": 22,
+    "r": 21,
+    "s": 20,
+    "t": 15,
+    "u": 26,
+    "v": 52,
+    "w": 27,
+    "y": 14,
+    "|": 4,
+    "à": 43,
+    "á": 16,
+    "è": 51,
+    "é": 41,
+    "ì": 49,
+    "í": 28,
+    "ò": 45,
+    "ó": 36,
+    "ù": 48,
+    "ú": 37,
+    "ě": 44,
+    "ń": 53,
+    "ŋ": 19,
+    "ǎ": 34,
+    "ǐ": 46,
+    "ǒ": 40,
+    "ǔ": 35,
+    "ɔ": 12,
+    "ɛ": 5,
+    "ɨ": 17,
+    "ʉ": 29,
+    "̀": 38,
+    "́": 11,
+    "̌": 30
+  },
+  "keo": {
+    "'": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 14,
+    "e": 11,
+    "f": 27,
+    "g": 16,
+    "h": 26,
+    "i": 6,
+    "k": 9,
+    "l": 12,
+    "m": 20,
+    "n": 10,
+    "o": 8,
+    "p": 21,
+    "r": 13,
+    "s": 22,
+    "t": 15,
+    "u": 7,
+    "v": 28,
+    "w": 25,
+    "y": 17,
+    "z": 24,
+    "|": 4,
+    "­": 29,
+    "ŋ": 19,
+    "—": 30
+  },
+  "ker": {
+    "'": 34,
+    "-": 30,
+    "0": 45,
+    "1": 44,
+    "2": 46,
+    "3": 47,
+    "4": 48,
+    "5": 49,
+    "6": 50,
+    "7": 51,
+    "8": 53,
+    "9": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 28,
+    "d": 16,
+    "e": 8,
+    "f": 29,
+    "g": 22,
+    "h": 25,
+    "i": 7,
+    "j": 27,
+    "k": 10,
+    "l": 18,
+    "m": 12,
+    "n": 11,
+    "o": 19,
+    "p": 24,
+    "r": 15,
+    "s": 21,
+    "t": 13,
+    "u": 14,
+    "v": 35,
+    "w": 20,
+    "y": 23,
+    "z": 31,
+    "|": 4,
+    "á": 32,
+    "ã": 36,
+    "é": 38,
+    "í": 52,
+    "ó": 37,
+    "õ": 39,
+    "ú": 41,
+    "ŋ": 9,
+    "ɓ": 33,
+    "ɗ": 26,
+    "ə": 6,
+    "̃": 42,
+    "ẽ": 40,
+    "–": 43
+  },
+  "key": {
+    "'": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ఁ": 33,
+    "ం": 24,
+    "అ": 29,
+    "ఆ": 34,
+    "ఇ": 40,
+    "ఈ": 41,
+    "ఉ": 38,
+    "ఊ": 49,
+    "ఎ": 37,
+    "ఏ": 42,
+    "ఐ": 48,
+    "ఒ": 39,
+    "ఓ": 45,
+    "ఔ": 53,
+    "క": 9,
+    "గ": 23,
+    "ఙ": 43,
+    "చ": 17,
+    "జ": 16,
+    "ట": 32,
+    "డ": 28,
+    "త": 10,
+    "ద": 18,
+    "న": 14,
+    "ప": 22,
+    "ఫ": 52,
+    "బ": 31,
+    "మ": 12,
+    "య": 21,
+    "ర": 15,
+    "ల": 13,
+    "వ": 26,
+    "శ": 54,
+    "ష": 50,
+    "స": 11,
+    "హ": 44,
+    "ా": 19,
+    "ి": 6,
+    "ీ": 36,
+    "ు": 8,
+    "ూ": 35,
+    "ె": 7,
+    "ే": 27,
+    "ై": 47,
+    "ొ": 20,
+    "ో": 30,
+    "ౌ": 51,
+    "్": 5,
+    "‍": 25
+  },
+  "kez": {
+    "'": 36,
+    "-": 27,
+    "0": 45,
+    "5": 50,
+    "6": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 32,
+    "d": 25,
+    "e": 17,
+    "f": 24,
+    "g": 13,
+    "h": 31,
+    "i": 14,
+    "j": 29,
+    "k": 9,
+    "l": 10,
+    "m": 15,
+    "n": 7,
+    "o": 22,
+    "p": 21,
+    "q": 43,
+    "r": 30,
+    "s": 19,
+    "t": 12,
+    "u": 16,
+    "v": 26,
+    "w": 20,
+    "x": 37,
+    "y": 18,
+    "z": 23,
+    "|": 4,
+    "à": 34,
+    "á": 35,
+    "è": 42,
+    "é": 38,
+    "ì": 41,
+    "í": 48,
+    "ù": 40,
+    "ú": 39,
+    "ń": 52,
+    "ǹ": 49,
+    "ɔ": 6,
+    "ɛ": 8,
+    "ɜ": 51,
+    "̀": 28,
+    "́": 33,
+    "–": 44,
+    "—": 47
+  },
+  "kfb": {
+    "'": 44,
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 19,
+    "अ": 41,
+    "आ": 17,
+    "इ": 28,
+    "ई": 37,
+    "उ": 38,
+    "ऊ": 43,
+    "ए": 31,
+    "ओ": 34,
+    "क": 10,
+    "ग": 18,
+    "च": 39,
+    "ज": 33,
+    "ट": 25,
+    "ड": 29,
+    "त": 9,
+    "द": 13,
+    "न": 6,
+    "प": 23,
+    "ब": 32,
+    "म": 16,
+    "य": 26,
+    "र": 12,
+    "ल": 20,
+    "ळ": 30,
+    "व": 22,
+    "श": 35,
+    "स": 21,
+    "ा": 5,
+    "ि": 27,
+    "ी": 11,
+    "ु": 14,
+    "ू": 36,
+    "ृ": 45,
+    "े": 8,
+    "ै": 40,
+    "ो": 24,
+    "ौ": 42,
+    "्": 7,
+    "०": 47,
+    "१": 49,
+    "२": 51,
+    "४": 48,
+    "६": 50,
+    "७": 52,
+    "‍": 15
+  },
+  "kff-script_telugu": {
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ం": 13,
+    "అ": 31,
+    "ఆ": 35,
+    "ఇ": 33,
+    "ఈ": 37,
+    "ఉ": 42,
+    "ఊ": 39,
+    "ఎ": 41,
+    "ఏ": 40,
+    "ఒ": 38,
+    "ఓ": 30,
+    "క": 11,
+    "గ": 19,
+    "చ": 34,
+    "జ": 29,
+    "ట": 22,
+    "డ": 24,
+    "త": 9,
+    "ద": 16,
+    "న": 6,
+    "ప": 20,
+    "బ": 27,
+    "మ": 15,
+    "య": 25,
+    "ర": 12,
+    "ల": 21,
+    "వ": 23,
+    "స": 17,
+    "ా": 10,
+    "ి": 5,
+    "ీ": 32,
+    "ు": 8,
+    "ూ": 36,
+    "ె": 14,
+    "ే": 26,
+    "ొ": 28,
+    "ో": 18,
+    "్": 7
+  },
+  "kfw": {
+    "'": 24,
+    "-": 22,
+    "0": 36,
+    "1": 29,
+    "2": 30,
+    "3": 31,
+    "4": 32,
+    "5": 33,
+    "6": 34,
+    "7": 35,
+    "8": 38,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 20,
+    "d": 23,
+    "e": 11,
+    "f": 40,
+    "g": 13,
+    "h": 9,
+    "i": 6,
+    "j": 21,
+    "k": 8,
+    "l": 16,
+    "m": 10,
+    "n": 7,
+    "o": 15,
+    "p": 19,
+    "q": 42,
+    "r": 17,
+    "s": 18,
+    "t": 14,
+    "u": 12,
+    "v": 39,
+    "w": 25,
+    "x": 41,
+    "y": 26,
+    "z": 28,
+    "|": 5,
+    "–": 44,
+    "—": 43
+  },
+  "kfx": {
+    "'": 65,
+    "-": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 38,
+    "ं": 21,
+    "ः": 61,
+    "अ": 51,
+    "आ": 30,
+    "इ": 43,
+    "ई": 34,
+    "उ": 58,
+    "ऊ": 44,
+    "ऋ": 64,
+    "ए": 46,
+    "ऐ": 37,
+    "ऑ": 59,
+    "ओ": 60,
+    "औ": 57,
+    "क": 12,
+    "ख": 41,
+    "ग": 33,
+    "घ": 45,
+    "च": 40,
+    "छ": 49,
+    "ज": 25,
+    "झ": 54,
+    "ञ": 62,
+    "ट": 48,
+    "ठ": 53,
+    "ड": 42,
+    "ढ": 47,
+    "ण": 26,
+    "त": 10,
+    "थ": 50,
+    "द": 27,
+    "ध": 39,
+    "न": 13,
+    "प": 24,
+    "फ": 52,
+    "ब": 15,
+    "भ": 36,
+    "म": 22,
+    "य": 28,
+    "र": 7,
+    "ल": 23,
+    "व": 35,
+    "श": 29,
+    "ष": 55,
+    "स": 9,
+    "ह": 11,
+    "़": 19,
+    "ा": 5,
+    "ि": 14,
+    "ी": 8,
+    "ु": 20,
+    "ू": 31,
+    "ृ": 63,
+    "े": 17,
+    "ै": 6,
+    "ो": 18,
+    "ौ": 32,
+    "्": 16,
+    "‍": 67,
+    "–": 66
+  },
+  "khg": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "་": 4,
+    "ཀ": 23,
+    "ཁ": 21,
+    "ག": 8,
+    "ང": 11,
+    "ཅ": 26,
+    "ཆ": 29,
+    "ཇ": 45,
+    "ཉ": 35,
+    "ཊ": 59,
+    "ཏ": 34,
+    "ཐ": 30,
+    "ད": 7,
+    "ན": 10,
+    "པ": 20,
+    "ཕ": 27,
+    "བ": 12,
+    "མ": 13,
+    "ཙ": 42,
+    "ཚ": 37,
+    "ཛ": 47,
+    "ཝ": 58,
+    "ཞ": 25,
+    "ཟ": 38,
+    "འ": 18,
+    "ཡ": 22,
+    "ར": 14,
+    "ལ": 19,
+    "ཤ": 32,
+    "ས": 5,
+    "ཧ": 51,
+    "ཨ": 49,
+    "ཪ": 60,
+    "ཱ": 39,
+    "ི": 6,
+    "ུ": 15,
+    "ེ": 16,
+    "ོ": 9,
+    "ྐ": 33,
+    "ྒ": 36,
+    "ྔ": 48,
+    "ྕ": 56,
+    "ྗ": 46,
+    "ྙ": 52,
+    "ྟ": 31,
+    "ྡ": 41,
+    "ྣ": 28,
+    "ྤ": 44,
+    "ྦ": 54,
+    "ྨ": 43,
+    "ྩ": 50,
+    "ྫ": 55,
+    "ྭ": 57,
+    "ྱ": 17,
+    "ྲ": 24,
+    "ླ": 40,
+    "ྷ": 53
+  },
+  "khm": {
+    "!": 140,
+    "\"": 118,
+    "$": 127,
+    "%": 119,
+    "&": 132,
+    "'": 113,
+    ",": 109,
+    "-": 106,
+    ".": 107,
+    "/": 115,
+    "0": 66,
+    "1": 73,
+    "2": 84,
+    "3": 100,
+    "4": 96,
+    "5": 92,
+    "6": 95,
+    "7": 104,
+    "8": 98,
+    "9": 88,
+    ":": 121,
+    ";": 133,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 141,
+    "[": 146,
+    "]": 147,
+    "a": 44,
+    "b": 81,
+    "c": 68,
+    "d": 71,
+    "e": 49,
+    "f": 90,
+    "g": 74,
+    "h": 67,
+    "i": 55,
+    "j": 101,
+    "k": 82,
+    "l": 60,
+    "m": 70,
+    "n": 53,
+    "o": 52,
+    "p": 75,
+    "q": 110,
+    "r": 51,
+    "s": 57,
+    "t": 61,
+    "u": 69,
+    "v": 94,
+    "w": 89,
+    "x": 108,
+    "y": 86,
+    "z": 105,
+    "£": 149,
+    "«": 124,
+    "°": 143,
+    "²": 148,
+    "»": 125,
+    "á": 128,
+    "ç": 126,
+    "é": 142,
+    "í": 136,
+    "õ": 138,
+    "ú": 150,
+    "ü": 122,
+    "ក": 8,
+    "ខ": 41,
+    "គ": 30,
+    "ឃ": 85,
+    "ង": 12,
+    "ច": 24,
+    "ឆ": 63,
+    "ជ": 29,
+    "ឈ": 76,
+    "ញ": 39,
+    "ដ": 18,
+    "ឋ": 72,
+    "ឌ": 83,
+    "ឍ": 111,
+    "ណ": 38,
+    "ត": 14,
+    "ថ": 42,
+    "ទ": 16,
+    "ធ": 40,
+    "ន": 6,
+    "ប": 9,
+    "ផ": 43,
+    "ព": 22,
+    "ភ": 45,
+    "ម": 11,
+    "យ": 17,
+    "រ": 7,
+    "ល": 13,
+    "វ": 26,
+    "ស": 10,
+    "ហ": 37,
+    "ឡ": 54,
+    "អ": 32,
+    "ឥ": 102,
+    "ឦ": 135,
+    "ឧ": 99,
+    "ឪ": 123,
+    "ឫ": 134,
+    "ឬ": 80,
+    "ឭ": 151,
+    "ឮ": 130,
+    "ឯ": 97,
+    "ឱ": 77,
+    "ឲ": 103,
+    "ឳ": 137,
+    "ា": 4,
+    "ិ": 19,
+    "ី": 25,
+    "ឹ": 47,
+    "ឺ": 56,
+    "ុ": 20,
+    "ូ": 31,
+    "ួ": 36,
+    "ើ": 27,
+    "ឿ": 87,
+    "ៀ": 64,
+    "េ": 21,
+    "ែ": 28,
+    "ៃ": 50,
+    "ោ": 33,
+    "ៅ": 35,
+    "ំ": 23,
+    "ះ": 34,
+    "ៈ": 78,
+    "៉": 46,
+    "៊": 59,
+    "់": 15,
+    "៌": 93,
+    "៍": 62,
+    "៎": 139,
+    "៏": 65,
+    "័": 58,
+    "្": 5,
+    "។": 91,
+    "៖": 112,
+    "ៗ": 79,
+    "០": 120,
+    "១": 117,
+    "២": 129,
+    "៣": 145,
+    "៥": 144,
+    "៧": 114,
+    "​": 48,
+    "—": 131,
+    "”": 116
+  },
+  "khq": {
+    "'": 31,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 24,
+    "d": 9,
+    "e": 11,
+    "f": 25,
+    "g": 16,
+    "h": 19,
+    "i": 6,
+    "j": 23,
+    "k": 12,
+    "l": 20,
+    "m": 15,
+    "n": 8,
+    "o": 7,
+    "p": 29,
+    "r": 13,
+    "s": 17,
+    "t": 22,
+    "u": 14,
+    "w": 21,
+    "y": 10,
+    "z": 28,
+    "|": 4,
+    "ã": 34,
+    "õ": 36,
+    "ĩ": 38,
+    "ŋ": 30,
+    "š": 32,
+    "ž": 33,
+    "ɲ": 27,
+    "ẽ": 35,
+    "–": 37
+  },
+  "kia": {
+    "'": 39,
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 35,
+    "d": 16,
+    "e": 6,
+    "f": 29,
+    "g": 9,
+    "h": 28,
+    "i": 12,
+    "j": 27,
+    "k": 11,
+    "l": 15,
+    "m": 8,
+    "n": 7,
+    "o": 10,
+    "p": 23,
+    "r": 17,
+    "s": 19,
+    "t": 18,
+    "u": 21,
+    "v": 30,
+    "w": 20,
+    "y": 14,
+    "z": 26,
+    "|": 4,
+    "à": 40,
+    "á": 31,
+    "è": 46,
+    "é": 37,
+    "í": 45,
+    "ò": 47,
+    "ó": 38,
+    "ù": 48,
+    "ú": 43,
+    "ŋ": 22,
+    "ū": 44,
+    "ɓ": 32,
+    "ɗ": 25,
+    "́": 42,
+    "̰": 24,
+    "ḛ": 33,
+    "ḭ": 34,
+    "ṵ": 36
+  },
+  "kij": {
+    "'": 25,
+    "-": 26,
+    "0": 28,
+    "1": 31,
+    "2": 34,
+    "4": 30,
+    "5": 35,
+    "6": 33,
+    "8": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 23,
+    "a": 4,
+    "b": 14,
+    "d": 19,
+    "e": 17,
+    "f": 36,
+    "g": 16,
+    "i": 6,
+    "k": 9,
+    "l": 7,
+    "m": 10,
+    "n": 18,
+    "o": 11,
+    "p": 22,
+    "r": 24,
+    "s": 13,
+    "t": 12,
+    "u": 8,
+    "v": 20,
+    "w": 15,
+    "y": 21,
+    "|": 5,
+    "á": 27,
+    "é": 32,
+    "ó": 29
+  },
+  "kik": {
+    "'": 30,
+    "-": 23,
+    "0": 33,
+    "1": 34,
+    "2": 36,
+    "4": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 22,
+    "d": 20,
+    "e": 14,
+    "f": 29,
+    "g": 15,
+    "h": 17,
+    "i": 10,
+    "j": 25,
+    "k": 16,
+    "l": 27,
+    "m": 13,
+    "n": 7,
+    "o": 11,
+    "p": 28,
+    "q": 38,
+    "r": 8,
+    "s": 26,
+    "t": 12,
+    "u": 19,
+    "v": 35,
+    "w": 18,
+    "y": 21,
+    "z": 31,
+    "|": 4,
+    "ĩ": 6,
+    "ũ": 9,
+    "ʼ": 32
+  },
+  "kin": {
+    "'": 24,
+    "-": 30,
+    "0": 140,
+    "1": 49,
+    "2": 62,
+    "3": 109,
+    "4": 57,
+    "5": 139,
+    "6": 136,
+    "7": 135,
+    "8": 138,
+    "9": 137,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "\\": 53,
+    "`": 36,
+    "a": 5,
+    "b": 11,
+    "c": 23,
+    "d": 22,
+    "e": 8,
+    "f": 26,
+    "g": 16,
+    "h": 19,
+    "i": 6,
+    "j": 25,
+    "k": 15,
+    "l": 29,
+    "m": 13,
+    "n": 9,
+    "o": 12,
+    "p": 28,
+    "q": 32,
+    "r": 10,
+    "s": 20,
+    "t": 18,
+    "u": 7,
+    "v": 27,
+    "w": 17,
+    "x": 31,
+    "y": 14,
+    "z": 21,
+    "|": 4,
+    "¤": 134,
+    "®": 94,
+    "ß": 124,
+    "à": 60,
+    "á": 37,
+    "â": 39,
+    "ã": 54,
+    "ä": 102,
+    "å": 118,
+    "æ": 76,
+    "ç": 41,
+    "è": 35,
+    "é": 33,
+    "ê": 55,
+    "ë": 51,
+    "ì": 73,
+    "í": 43,
+    "î": 66,
+    "ï": 47,
+    "ð": 125,
+    "ñ": 63,
+    "ò": 89,
+    "ó": 52,
+    "ô": 42,
+    "ö": 64,
+    "ø": 98,
+    "ù": 77,
+    "ú": 46,
+    "û": 71,
+    "ü": 58,
+    "ÿ": 93,
+    "ā": 38,
+    "ă": 128,
+    "ć": 72,
+    "č": 79,
+    "ē": 59,
+    "ę": 106,
+    "ğ": 83,
+    "ī": 45,
+    "ł": 99,
+    "ō": 44,
+    "œ": 68,
+    "ş": 81,
+    "š": 87,
+    "ū": 56,
+    "ż": 132,
+    "ž": 115,
+    "ȃ": 101,
+    "ȋ": 116,
+    "ȏ": 117,
+    "ʺ": 80,
+    "ʻ": 96,
+    "ʼ": 65,
+    "ʽ": 67,
+    "ˮ": 82,
+    "́": 61,
+    "̃": 91,
+    "̄": 105,
+    "̈": 78,
+    "̋": 88,
+    "̧": 126,
+    "а": 69,
+    "б": 103,
+    "в": 86,
+    "д": 120,
+    "е": 85,
+    "ж": 121,
+    "и": 104,
+    "к": 74,
+    "м": 127,
+    "н": 97,
+    "о": 114,
+    "р": 90,
+    "с": 108,
+    "т": 122,
+    "у": 84,
+    "ф": 133,
+    "ѐ": 110,
+    "і": 92,
+    "ҫ": 111,
+    "ӱ": 123,
+    "ب": 129,
+    "ش": 131,
+    "ە": 130,
+    "ḍ": 107,
+    "ḥ": 70,
+    "ṭ": 113,
+    "ẓ": 100,
+    "ị": 112,
+    "‐": 75,
+    "–": 34,
+    "—": 50,
+    "′": 48,
+    "‽": 40,
+    "√": 95,
+    "❤": 119
+  },
+  "kir": {
+    "\"": 83,
+    "%": 81,
+    "&": 95,
+    "'": 72,
+    "*": 97,
+    "+": 89,
+    ",": 71,
+    "-": 33,
+    ".": 56,
+    "/": 70,
+    "0": 35,
+    "1": 38,
+    "2": 40,
+    "3": 52,
+    "4": 48,
+    "5": 45,
+    "6": 50,
+    "7": 58,
+    "8": 55,
+    "9": 43,
+    ":": 66,
+    ";": 91,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 98,
+    "]": 99,
+    "a": 42,
+    "b": 67,
+    "c": 57,
+    "d": 63,
+    "e": 44,
+    "f": 68,
+    "g": 61,
+    "h": 64,
+    "i": 49,
+    "j": 82,
+    "k": 69,
+    "l": 59,
+    "m": 65,
+    "n": 47,
+    "o": 46,
+    "p": 60,
+    "q": 85,
+    "r": 51,
+    "s": 54,
+    "t": 53,
+    "u": 62,
+    "v": 75,
+    "w": 78,
+    "x": 84,
+    "y": 77,
+    "z": 86,
+    "|": 4,
+    "¥": 88,
+    "°": 92,
+    "²": 96,
+    "õ": 87,
+    "ú": 100,
+    "́": 94,
+    "а": 5,
+    "б": 19,
+    "в": 32,
+    "г": 18,
+    "д": 15,
+    "е": 11,
+    "ж": 23,
+    "з": 28,
+    "и": 13,
+    "й": 26,
+    "к": 9,
+    "л": 10,
+    "м": 20,
+    "н": 6,
+    "о": 16,
+    "п": 25,
+    "р": 7,
+    "с": 21,
+    "т": 8,
+    "у": 14,
+    "ф": 34,
+    "х": 39,
+    "ц": 37,
+    "ч": 27,
+    "ш": 24,
+    "щ": 93,
+    "ъ": 76,
+    "ы": 12,
+    "ь": 41,
+    "э": 29,
+    "ю": 36,
+    "я": 30,
+    "ё": 73,
+    "ң": 31,
+    "ү": 17,
+    "ө": 22,
+    "–": 79,
+    "—": 74,
+    "’": 80,
+    "№": 90
+  },
+  "kjb": {
+    "'": 14,
+    "-": 37,
+    "0": 40,
+    "1": 41,
+    "2": 42,
+    "3": 45,
+    "4": 43,
+    "5": 44,
+    "6": 47,
+    "7": 48,
+    "8": 49,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 9,
+    "d": 27,
+    "e": 11,
+    "f": 36,
+    "g": 31,
+    "h": 21,
+    "i": 6,
+    "j": 15,
+    "k": 19,
+    "l": 10,
+    "m": 20,
+    "n": 7,
+    "o": 13,
+    "p": 24,
+    "q": 29,
+    "r": 26,
+    "s": 22,
+    "t": 8,
+    "u": 16,
+    "v": 39,
+    "w": 23,
+    "x": 18,
+    "y": 12,
+    "z": 25,
+    "|": 4,
+    "á": 35,
+    "é": 34,
+    "í": 33,
+    "ñ": 50,
+    "ó": 38,
+    "ú": 30,
+    "ẍ": 28,
+    "—": 32
+  },
+  "kje": {
+    "'": 16,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 29,
+    "d": 17,
+    "e": 8,
+    "f": 27,
+    "g": 25,
+    "h": 14,
+    "i": 7,
+    "j": 28,
+    "k": 13,
+    "l": 12,
+    "m": 10,
+    "n": 6,
+    "o": 9,
+    "p": 19,
+    "r": 11,
+    "s": 20,
+    "t": 22,
+    "u": 15,
+    "w": 18,
+    "y": 21,
+    "z": 26,
+    "|": 4
+  },
+  "kjg": {
+    "'": 7,
+    "-": 32,
+    "0": 36,
+    "2": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "c": 25,
+    "d": 27,
+    "e": 22,
+    "f": 34,
+    "g": 9,
+    "h": 8,
+    "i": 16,
+    "j": 31,
+    "k": 18,
+    "l": 17,
+    "m": 11,
+    "n": 6,
+    "o": 10,
+    "p": 24,
+    "r": 13,
+    "s": 21,
+    "t": 15,
+    "u": 20,
+    "v": 29,
+    "w": 19,
+    "y": 12,
+    "|": 4,
+    "è": 26,
+    "é": 23,
+    "ê": 33,
+    "ñ": 30,
+    "ô": 14
+  },
+  "kjh": {
+    "-": 32,
+    "0": 47,
+    "1": 44,
+    "2": 45,
+    "4": 46,
+    "6": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "́": 43,
+    "а": 5,
+    "б": 27,
+    "в": 34,
+    "г": 28,
+    "д": 19,
+    "е": 9,
+    "з": 20,
+    "и": 15,
+    "й": 26,
+    "к": 22,
+    "л": 12,
+    "м": 25,
+    "н": 6,
+    "о": 17,
+    "п": 14,
+    "р": 7,
+    "с": 11,
+    "т": 13,
+    "у": 24,
+    "ф": 35,
+    "х": 16,
+    "ц": 41,
+    "ч": 18,
+    "ш": 42,
+    "ы": 8,
+    "ь": 37,
+    "э": 36,
+    "ю": 40,
+    "я": 39,
+    "ё": 38,
+    "і": 10,
+    "ғ": 23,
+    "ң": 21,
+    "ҷ": 31,
+    "ӧ": 30,
+    "ӱ": 29,
+    "–": 33
+  },
+  "kki": {
+    "'": 26,
+    "-": 30,
+    "1": 31,
+    "2": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 22,
+    "d": 19,
+    "e": 9,
+    "f": 20,
+    "g": 16,
+    "h": 15,
+    "i": 6,
+    "j": 24,
+    "k": 12,
+    "l": 14,
+    "m": 13,
+    "n": 8,
+    "o": 10,
+    "p": 25,
+    "q": 29,
+    "s": 18,
+    "t": 21,
+    "u": 7,
+    "v": 27,
+    "w": 11,
+    "y": 17,
+    "|": 4
+  },
+  "kkj": {
+    "'": 35,
+    "-": 43,
+    "1": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "d": 21,
+    "e": 8,
+    "f": 39,
+    "g": 26,
+    "h": 31,
+    "i": 12,
+    "j": 24,
+    "k": 10,
+    "l": 23,
+    "m": 9,
+    "n": 7,
+    "o": 18,
+    "p": 27,
+    "r": 30,
+    "s": 17,
+    "t": 13,
+    "u": 16,
+    "v": 40,
+    "w": 19,
+    "y": 11,
+    "|": 4,
+    "à": 28,
+    "á": 5,
+    "ã": 29,
+    "ä": 15,
+    "è": 34,
+    "ê": 33,
+    "ë": 25,
+    "ì": 38,
+    "î": 37,
+    "ï": 32,
+    "ñ": 20,
+    "ó": 14,
+    "õ": 41,
+    "ù": 36
+  },
+  "kle": {
+    "'": 57,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 53,
+    "ं": 54,
+    "ः": 41,
+    "अ": 38,
+    "आ": 26,
+    "इ": 49,
+    "ई": 24,
+    "उ": 55,
+    "ऊ": 31,
+    "ए": 45,
+    "ऐ": 62,
+    "ओ": 34,
+    "क": 10,
+    "ख": 19,
+    "ग": 32,
+    "घ": 58,
+    "ङ": 18,
+    "च": 22,
+    "छ": 28,
+    "ज": 36,
+    "झ": 61,
+    "ञ": 51,
+    "ट": 40,
+    "ठ": 59,
+    "ड": 47,
+    "ढ": 64,
+    "ण": 48,
+    "त": 17,
+    "थ": 37,
+    "द": 25,
+    "ध": 46,
+    "न": 11,
+    "प": 12,
+    "फ": 44,
+    "ब": 30,
+    "भ": 52,
+    "म": 8,
+    "य": 23,
+    "र": 21,
+    "ल": 16,
+    "व": 27,
+    "श": 35,
+    "ष": 42,
+    "स": 15,
+    "ह": 29,
+    "ा": 6,
+    "ि": 33,
+    "ी": 7,
+    "ु": 43,
+    "ू": 13,
+    "ृ": 63,
+    "े": 20,
+    "ै": 50,
+    "ो": 9,
+    "ौ": 56,
+    "्": 5,
+    "‍": 14,
+    "–": 60
+  },
+  "klu": {
+    "'": 50,
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 18,
+    "c": 42,
+    "d": 27,
+    "e": 8,
+    "f": 38,
+    "g": 40,
+    "h": 22,
+    "i": 13,
+    "j": 23,
+    "k": 14,
+    "l": 10,
+    "m": 21,
+    "n": 6,
+    "o": 12,
+    "p": 25,
+    "s": 19,
+    "t": 16,
+    "u": 20,
+    "w": 24,
+    "y": 17,
+    "|": 4,
+    "à": 35,
+    "â": 28,
+    "è": 29,
+    "ê": 30,
+    "ì": 36,
+    "î": 37,
+    "ò": 33,
+    "ô": 32,
+    "ù": 39,
+    "û": 41,
+    "ě": 47,
+    "ǎ": 49,
+    "ǐ": 48,
+    "ǒ": 52,
+    "ǔ": 45,
+    "ɔ": 11,
+    "ɛ": 9,
+    "ˆ": 43,
+    "ˈ": 44,
+    "˖": 31,
+    "̀": 26,
+    "̂": 34,
+    "̌": 51,
+    "̍": 5,
+    "‐": 15
+  },
+  "klv": {
+    "'": 33,
+    "-": 26,
+    "0": 32,
+    "1": 34,
+    "2": 35,
+    "3": 38,
+    "4": 36,
+    "5": 37,
+    "6": 39,
+    "7": 41,
+    "8": 43,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 23,
+    "e": 11,
+    "f": 31,
+    "g": 17,
+    "h": 16,
+    "i": 7,
+    "j": 30,
+    "k": 21,
+    "l": 9,
+    "m": 15,
+    "n": 6,
+    "o": 10,
+    "p": 24,
+    "r": 19,
+    "s": 13,
+    "t": 8,
+    "u": 14,
+    "v": 20,
+    "w": 29,
+    "y": 28,
+    "|": 4,
+    "ŋ": 25,
+    "ǝ": 12,
+    "̃": 22,
+    "ṽ": 27,
+    "—": 40
+  },
+  "klw": {
+    "'": 27,
+    "-": 19,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 24,
+    "d": 18,
+    "e": 10,
+    "f": 28,
+    "g": 21,
+    "h": 26,
+    "i": 8,
+    "j": 25,
+    "k": 16,
+    "l": 17,
+    "m": 11,
+    "n": 6,
+    "o": 7,
+    "p": 13,
+    "r": 14,
+    "s": 15,
+    "t": 12,
+    "u": 9,
+    "w": 23,
+    "y": 22,
+    "z": 29,
+    "|": 5
+  },
+  "kma": {
+    "'": 34,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 30,
+    "d": 11,
+    "e": 10,
+    "f": 31,
+    "g": 19,
+    "h": 26,
+    "i": 7,
+    "j": 29,
+    "k": 16,
+    "l": 21,
+    "m": 12,
+    "n": 9,
+    "o": 23,
+    "p": 28,
+    "r": 24,
+    "s": 18,
+    "t": 15,
+    "u": 22,
+    "v": 25,
+    "w": 20,
+    "y": 17,
+    "z": 32,
+    "|": 4,
+    "ŋ": 8,
+    "ɔ": 27,
+    "ɩ": 6,
+    "ʊ": 14
+  },
+  "kmd": {
+    "'": 33,
+    "-": 25,
+    "0": 35,
+    "1": 38,
+    "4": 37,
+    "6": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 26,
+    "d": 15,
+    "e": 10,
+    "f": 31,
+    "g": 12,
+    "h": 29,
+    "i": 11,
+    "j": 21,
+    "k": 16,
+    "l": 20,
+    "m": 13,
+    "n": 6,
+    "o": 18,
+    "p": 17,
+    "q": 36,
+    "r": 19,
+    "s": 9,
+    "t": 14,
+    "u": 7,
+    "v": 24,
+    "w": 22,
+    "x": 34,
+    "y": 8,
+    "z": 32,
+    "|": 5,
+    "é": 30,
+    "ʼ": 28,
+    "ꞌ": 27
+  },
+  "kml": {
+    "'": 23,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 29,
+    "d": 9,
+    "e": 15,
+    "f": 30,
+    "g": 13,
+    "h": 28,
+    "i": 8,
+    "j": 26,
+    "k": 27,
+    "l": 20,
+    "m": 16,
+    "n": 6,
+    "o": 14,
+    "p": 18,
+    "q": 34,
+    "r": 24,
+    "s": 12,
+    "t": 11,
+    "u": 7,
+    "v": 32,
+    "w": 19,
+    "x": 33,
+    "y": 17,
+    "z": 31,
+    "|": 5,
+    "ḵ": 10,
+    "ḻ": 22
+  },
+  "kmr-script_arabic": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "b": 39,
+    "c": 40,
+    "m": 38,
+    "|": 4,
+    "ئ": 20,
+    "ا": 6,
+    "ب": 13,
+    "ت": 15,
+    "ج": 30,
+    "ح": 33,
+    "خ": 23,
+    "د": 12,
+    "ر": 11,
+    "ز": 26,
+    "س": 19,
+    "ش": 27,
+    "ع": 35,
+    "غ": 37,
+    "ف": 32,
+    "ق": 34,
+    "ك": 14,
+    "ل": 18,
+    "م": 16,
+    "ن": 8,
+    "ه": 17,
+    "و": 9,
+    "پ": 28,
+    "چ": 29,
+    "ڕ": 31,
+    "ژ": 25,
+    "ڤ": 22,
+    "گ": 24,
+    "ڵ": 36,
+    "ۆ": 21,
+    "ی": 7,
+    "ێ": 10,
+    "ە": 5
+  },
+  "kmr-script_cyrillic": {
+    "'": 16,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "q": 34,
+    "w": 15,
+    "|": 4,
+    "а": 6,
+    "б": 14,
+    "в": 26,
+    "г": 27,
+    "д": 12,
+    "е": 9,
+    "ж": 28,
+    "з": 30,
+    "и": 11,
+    "й": 19,
+    "к": 13,
+    "л": 24,
+    "м": 18,
+    "н": 8,
+    "о": 29,
+    "п": 32,
+    "р": 10,
+    "с": 22,
+    "т": 17,
+    "у": 20,
+    "ф": 36,
+    "х": 23,
+    "ч": 33,
+    "ш": 31,
+    "щ": 35,
+    "ь": 7,
+    "һ": 21,
+    "ә": 5,
+    "ӧ": 25
+  },
+  "kmr-script_latin": {
+    "'": 36,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 13,
+    "c": 33,
+    "d": 11,
+    "e": 5,
+    "f": 35,
+    "g": 28,
+    "h": 20,
+    "i": 6,
+    "j": 24,
+    "k": 12,
+    "l": 18,
+    "m": 15,
+    "n": 8,
+    "o": 25,
+    "p": 31,
+    "q": 34,
+    "r": 9,
+    "s": 21,
+    "t": 16,
+    "u": 30,
+    "v": 22,
+    "w": 17,
+    "x": 26,
+    "y": 19,
+    "z": 27,
+    "|": 4,
+    "ç": 32,
+    "ê": 10,
+    "î": 14,
+    "û": 23,
+    "ğ": 38,
+    "ş": 29,
+    "–": 39
+  },
+  "kmu": {
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "e": 6,
+    "f": 21,
+    "g": 14,
+    "h": 11,
+    "i": 8,
+    "k": 16,
+    "l": 18,
+    "m": 9,
+    "n": 7,
+    "o": 10,
+    "p": 22,
+    "s": 20,
+    "t": 15,
+    "u": 13,
+    "v": 19,
+    "y": 12,
+    "|": 5,
+    "ꞌ": 17
+  },
+  "knb": {
+    "'": 44,
+    "-": 26,
+    "0": 37,
+    "1": 40,
+    "2": 42,
+    "4": 39,
+    "6": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 14,
+    "d": 22,
+    "e": 20,
+    "f": 34,
+    "g": 15,
+    "h": 11,
+    "i": 7,
+    "j": 28,
+    "k": 12,
+    "l": 9,
+    "m": 13,
+    "n": 8,
+    "o": 18,
+    "p": 19,
+    "q": 38,
+    "r": 27,
+    "s": 10,
+    "t": 16,
+    "u": 6,
+    "v": 32,
+    "w": 23,
+    "x": 35,
+    "y": 17,
+    "z": 30,
+    "|": 4,
+    "à": 31,
+    "è": 43,
+    "ì": 36,
+    "ò": 33,
+    "ù": 29,
+    "ɏ": 24,
+    "ʼ": 25
+  },
+  "kne": {
+    "'": 26,
+    "-": 21,
+    "0": 34,
+    "1": 35,
+    "3": 33,
+    "4": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 25,
+    "d": 8,
+    "e": 12,
+    "f": 28,
+    "g": 14,
+    "h": 27,
+    "i": 7,
+    "j": 24,
+    "k": 16,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 9,
+    "p": 18,
+    "q": 32,
+    "r": 22,
+    "s": 11,
+    "t": 15,
+    "u": 23,
+    "v": 30,
+    "w": 20,
+    "x": 31,
+    "y": 10,
+    "z": 29,
+    "|": 4
+  },
+  "knf": {
+    "'": 31,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 34,
+    "d": 20,
+    "e": 15,
+    "f": 29,
+    "g": 30,
+    "h": 26,
+    "i": 6,
+    "j": 21,
+    "k": 9,
+    "l": 16,
+    "m": 18,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "r": 25,
+    "t": 27,
+    "u": 8,
+    "w": 14,
+    "y": 24,
+    "|": 4,
+    "ë": 17,
+    "ñ": 28,
+    "ŋ": 12,
+    "ş": 22,
+    "ţ": 13,
+    "ŧ": 23,
+    "–": 33
+  },
+  "knj": {
+    "'": 12,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 11,
+    "d": 26,
+    "e": 6,
+    "f": 38,
+    "g": 32,
+    "h": 21,
+    "i": 7,
+    "j": 9,
+    "k": 23,
+    "l": 13,
+    "m": 20,
+    "n": 8,
+    "o": 14,
+    "p": 24,
+    "q": 29,
+    "r": 25,
+    "s": 18,
+    "t": 10,
+    "u": 16,
+    "v": 36,
+    "w": 22,
+    "x": 17,
+    "y": 15,
+    "z": 27,
+    "|": 4,
+    "á": 31,
+    "é": 35,
+    "í": 34,
+    "ó": 39,
+    "ú": 30,
+    "ẍ": 28,
+    "—": 33
+  },
+  "knk": {
+    "'": 29,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 28,
+    "d": 22,
+    "e": 16,
+    "f": 25,
+    "g": 23,
+    "h": 26,
+    "i": 7,
+    "k": 10,
+    "l": 9,
+    "m": 12,
+    "n": 6,
+    "o": 11,
+    "p": 27,
+    "r": 18,
+    "s": 21,
+    "t": 20,
+    "u": 15,
+    "w": 19,
+    "y": 14,
+    "|": 4,
+    "ɔ": 13,
+    "ɛ": 8
+  },
+  "kno": {
+    "'": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 14,
+    "d": 18,
+    "e": 11,
+    "f": 24,
+    "g": 25,
+    "h": 15,
+    "i": 9,
+    "k": 16,
+    "m": 7,
+    "n": 6,
+    "o": 13,
+    "p": 27,
+    "s": 17,
+    "t": 19,
+    "u": 20,
+    "w": 22,
+    "y": 21,
+    "|": 4,
+    "ɔ": 10,
+    "ɛ": 8,
+    "‐": 23
+  },
+  "kog": {
+    "'": 36,
+    "0": 35,
+    "1": 37,
+    "2": 38,
+    "3": 41,
+    "4": 39,
+    "5": 40,
+    "6": 43,
+    "7": 42,
+    "8": 44,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 47,
+    "d": 11,
+    "e": 9,
+    "g": 13,
+    "h": 15,
+    "i": 7,
+    "j": 19,
+    "k": 6,
+    "l": 12,
+    "m": 20,
+    "n": 10,
+    "o": 28,
+    "p": 31,
+    "s": 14,
+    "t": 18,
+    "u": 8,
+    "w": 26,
+    "x": 21,
+    "y": 27,
+    "z": 16,
+    "|": 5,
+    "á": 22,
+    "ã": 46,
+    "é": 25,
+    "í": 24,
+    "ñ": 17,
+    "ó": 33,
+    "ú": 30,
+    "ʉ": 29,
+    "́": 34,
+    "—": 32
+  },
+  "kor": {
+    "!": 863,
+    "\"": 41,
+    "$": 920,
+    "%": 536,
+    "'": 292,
+    "*": 1060,
+    ",": 339,
+    "-": 357,
+    ".": 99,
+    "/": 472,
+    "0": 50,
+    "1": 53,
+    "2": 88,
+    "3": 154,
+    "4": 163,
+    "5": 155,
+    "6": 173,
+    "7": 242,
+    "8": 194,
+    "9": 140,
+    ":": 782,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 709,
+    "[": 936,
+    "]": 937,
+    "a": 67,
+    "b": 321,
+    "c": 197,
+    "d": 198,
+    "e": 110,
+    "f": 371,
+    "g": 260,
+    "h": 215,
+    "i": 121,
+    "j": 484,
+    "k": 266,
+    "l": 169,
+    "m": 157,
+    "n": 125,
+    "o": 118,
+    "p": 199,
+    "q": 563,
+    "r": 133,
+    "s": 122,
+    "t": 147,
+    "u": 219,
+    "v": 332,
+    "w": 426,
+    "x": 510,
+    "y": 354,
+    "z": 551,
+    "|": 4,
+    "~": 591,
+    "°": 964,
+    "²": 1033,
+    "á": 1057,
+    "ç": 1015,
+    "õ": 833,
+    "ú": 955,
+    "ü": 842,
+    "​": 831,
+    "–": 931,
+    "”": 761,
+    "℃": 839,
+    "因": 892,
+    "死": 891,
+    "가": 15,
+    "각": 193,
+    "간": 120,
+    "갇": 876,
+    "갈": 359,
+    "감": 273,
+    "갑": 681,
+    "값": 620,
+    "갓": 1156,
+    "갔": 698,
+    "강": 223,
+    "갖": 482,
+    "갗": 1279,
+    "같": 239,
+    "갚": 1101,
+    "개": 108,
+    "객": 346,
+    "갤": 1043,
+    "갯": 1282,
+    "갱": 1042,
+    "갸": 1256,
+    "걀": 910,
+    "거": 91,
+    "걱": 731,
+    "건": 181,
+    "걷": 807,
+    "걸": 376,
+    "검": 454,
+    "겁": 514,
+    "것": 48,
+    "겅": 1219,
+    "겉": 983,
+    "게": 56,
+    "겐": 760,
+    "겔": 862,
+    "겜": 1240,
+    "겟": 1238,
+    "겠": 464,
+    "겨": 329,
+    "격": 248,
+    "겪": 704,
+    "견": 311,
+    "결": 180,
+    "겸": 1076,
+    "겹": 923,
+    "겼": 594,
+    "경": 79,
+    "곁": 1103,
+    "계": 105,
+    "고": 17,
+    "곡": 554,
+    "곤": 634,
+    "곧": 938,
+    "골": 393,
+    "곰": 721,
+    "곱": 565,
+    "곳": 299,
+    "공": 71,
+    "과": 49,
+    "곽": 1295,
+    "관": 97,
+    "괄": 881,
+    "광": 280,
+    "괜": 774,
+    "괴": 529,
+    "굉": 1090,
+    "교": 152,
+    "구": 66,
+    "국": 58,
+    "군": 228,
+    "굳": 1065,
+    "굴": 506,
+    "굵": 1268,
+    "굶": 1130,
+    "굽": 843,
+    "궁": 724,
+    "궂": 1232,
+    "권": 360,
+    "궤": 1168,
+    "귀": 436,
+    "귈": 1265,
+    "귐": 1172,
+    "귓": 1325,
+    "규": 441,
+    "균": 668,
+    "그": 39,
+    "극": 391,
+    "근": 261,
+    "글": 465,
+    "금": 240,
+    "급": 259,
+    "긋": 917,
+    "긍": 887,
+    "기": 18,
+    "긴": 372,
+    "길": 351,
+    "김": 1120,
+    "깁": 1124,
+    "깃": 658,
+    "깅": 769,
+    "깊": 792,
+    "까": 255,
+    "깍": 1281,
+    "깎": 901,
+    "깐": 1185,
+    "깔": 895,
+    "깜": 860,
+    "깝": 725,
+    "깥": 610,
+    "깨": 430,
+    "꺼": 598,
+    "꺾": 1188,
+    "껄": 1290,
+    "껏": 908,
+    "껑": 1316,
+    "께": 412,
+    "껴": 1236,
+    "꼈": 939,
+    "꼬": 1078,
+    "꼭": 577,
+    "꼴": 1056,
+    "꼽": 788,
+    "꼿": 975,
+    "꽂": 1292,
+    "꽃": 884,
+    "꽤": 824,
+    "꽹": 1259,
+    "꾀": 1146,
+    "꾐": 1243,
+    "꾸": 622,
+    "꾼": 888,
+    "꿀": 1210,
+    "꿇": 1123,
+    "꿈": 780,
+    "꿉": 1029,
+    "꿔": 1055,
+    "꿨": 921,
+    "꿰": 1195,
+    "뀌": 768,
+    "뀐": 1087,
+    "뀔": 859,
+    "끄": 500,
+    "끈": 720,
+    "끊": 1079,
+    "끌": 544,
+    "끔": 755,
+    "끗": 1097,
+    "끝": 421,
+    "끼": 471,
+    "낀": 1066,
+    "낄": 837,
+    "낌": 1047,
+    "나": 40,
+    "낙": 745,
+    "낚": 1192,
+    "난": 235,
+    "낟": 1323,
+    "날": 291,
+    "낡": 988,
+    "남": 208,
+    "납": 645,
+    "낫": 1105,
+    "났": 460,
+    "낭": 699,
+    "낮": 445,
+    "낯": 1227,
+    "낱": 1135,
+    "낳": 1098,
+    "내": 144,
+    "낸": 715,
+    "낼": 557,
+    "냄": 729,
+    "냅": 661,
+    "냈": 550,
+    "냉": 1002,
+    "냐": 587,
+    "냥": 540,
+    "너": 281,
+    "넉": 1122,
+    "넌": 1305,
+    "널": 528,
+    "넓": 692,
+    "넘": 457,
+    "넙": 1003,
+    "넛": 971,
+    "넜": 1239,
+    "넣": 669,
+    "네": 290,
+    "넨": 1040,
+    "넬": 1140,
+    "넷": 513,
+    "녀": 433,
+    "녁": 666,
+    "년": 104,
+    "념": 419,
+    "녔": 1151,
+    "녕": 1004,
+    "노": 222,
+    "녹": 853,
+    "논": 726,
+    "놀": 747,
+    "놈": 871,
+    "놋": 1230,
+    "농": 487,
+    "높": 420,
+    "놓": 642,
+    "뇌": 716,
+    "뇨": 783,
+    "누": 399,
+    "눈": 361,
+    "눌": 956,
+    "눔": 1247,
+    "눕": 1177,
+    "눠": 1296,
+    "뉘": 1133,
+    "뉩": 785,
+    "뉴": 378,
+    "늄": 827,
+    "늉": 1306,
+    "느": 491,
+    "는": 7,
+    "늘": 300,
+    "늙": 1167,
+    "능": 214,
+    "늦": 1182,
+    "늪": 1034,
+    "니": 11,
+    "닉": 996,
+    "닌": 499,
+    "닐": 1045,
+    "님": 677,
+    "닙": 667,
+    "닛": 1297,
+    "닝": 607,
+    "닢": 1166,
+    "다": 5,
+    "닥": 582,
+    "닦": 1149,
+    "단": 182,
+    "닫": 823,
+    "달": 202,
+    "닭": 1116,
+    "닮": 1287,
+    "담": 524,
+    "답": 518,
+    "닷": 976,
+    "당": 136,
+    "닻": 1187,
+    "닿": 766,
+    "대": 36,
+    "댁": 1267,
+    "댄": 980,
+    "댈": 1039,
+    "댑": 1284,
+    "댓": 872,
+    "댔": 1190,
+    "더": 137,
+    "덕": 635,
+    "던": 230,
+    "덜": 643,
+    "덟": 1000,
+    "덤": 527,
+    "덥": 1255,
+    "덧": 815,
+    "덩": 673,
+    "덫": 844,
+    "덮": 559,
+    "데": 162,
+    "덱": 1136,
+    "덴": 567,
+    "델": 744,
+    "뎅": 1202,
+    "도": 29,
+    "독": 227,
+    "돈": 678,
+    "돋": 1134,
+    "돌": 315,
+    "돔": 1142,
+    "돕": 808,
+    "돗": 919,
+    "동": 65,
+    "돛": 1329,
+    "돼": 756,
+    "됐": 573,
+    "되": 59,
+    "된": 106,
+    "될": 297,
+    "됨": 1025,
+    "됩": 333,
+    "됭": 1088,
+    "두": 150,
+    "둑": 1119,
+    "둔": 623,
+    "둘": 521,
+    "둠": 1109,
+    "둡": 972,
+    "둥": 750,
+    "뒤": 455,
+    "뒷": 948,
+    "뒹": 1321,
+    "듀": 1059,
+    "드": 89,
+    "득": 388,
+    "든": 224,
+    "듣": 1091,
+    "들": 30,
+    "듬": 898,
+    "듭": 835,
+    "듯": 732,
+    "등": 285,
+    "디": 229,
+    "딘": 654,
+    "딛": 930,
+    "딜": 812,
+    "딥": 1280,
+    "딩": 626,
+    "딪": 1214,
+    "따": 203,
+    "딱": 1068,
+    "딴": 899,
+    "딸": 992,
+    "땅": 672,
+    "땋": 1320,
+    "때": 123,
+    "떠": 415,
+    "떡": 1189,
+    "떤": 356,
+    "떨": 402,
+    "떱": 1242,
+    "떳": 1152,
+    "떴": 1258,
+    "떻": 628,
+    "떼": 990,
+    "뗄": 1181,
+    "또": 184,
+    "똑": 857,
+    "똥": 929,
+    "뚜": 1016,
+    "뚝": 1073,
+    "뚤": 1193,
+    "뚫": 986,
+    "뚱": 1307,
+    "뛰": 664,
+    "뜨": 533,
+    "뜩": 1206,
+    "뜬": 834,
+    "뜯": 1128,
+    "뜰": 740,
+    "뜸": 1155,
+    "뜻": 502,
+    "띄": 867,
+    "띠": 978,
+    "띤": 1262,
+    "라": 42,
+    "락": 411,
+    "란": 318,
+    "랄": 985,
+    "람": 128,
+    "랍": 613,
+    "랐": 601,
+    "랑": 301,
+    "래": 249,
+    "랙": 615,
+    "랜": 265,
+    "램": 580,
+    "랩": 810,
+    "랫": 718,
+    "랬": 905,
+    "랭": 1046,
+    "랴": 1102,
+    "략": 498,
+    "량": 366,
+    "러": 95,
+    "럭": 575,
+    "런": 338,
+    "럴": 757,
+    "럼": 350,
+    "럽": 320,
+    "럿": 1178,
+    "렀": 804,
+    "렁": 1064,
+    "렇": 509,
+    "레": 210,
+    "렉": 636,
+    "렌": 405,
+    "렐": 1180,
+    "렘": 829,
+    "렙": 994,
+    "렛": 927,
+    "렝": 977,
+    "려": 114,
+    "력": 195,
+    "련": 340,
+    "렬": 963,
+    "렴": 504,
+    "렵": 611,
+    "렷": 1017,
+    "렸": 458,
+    "령": 336,
+    "례": 452,
+    "로": 14,
+    "록": 165,
+    "론": 296,
+    "롤": 633,
+    "롬": 627,
+    "롭": 590,
+    "롯": 637,
+    "롱": 882,
+    "뢰": 705,
+    "뢸": 1302,
+    "료": 327,
+    "룟": 1197,
+    "룡": 885,
+    "루": 196,
+    "룩": 593,
+    "룬": 1208,
+    "룰": 1085,
+    "룸": 820,
+    "룹": 604,
+    "룻": 1084,
+    "룽": 880,
+    "뤄": 671,
+    "뤼": 1058,
+    "류": 238,
+    "륙": 466,
+    "륜": 690,
+    "률": 561,
+    "륨": 1041,
+    "륭": 728,
+    "르": 87,
+    "륵": 1319,
+    "른": 151,
+    "를": 23,
+    "름": 294,
+    "릅": 569,
+    "릇": 1100,
+    "릉": 973,
+    "릎": 942,
+    "리": 25,
+    "릭": 670,
+    "린": 246,
+    "릴": 408,
+    "림": 317,
+    "립": 310,
+    "릿": 989,
+    "링": 526,
+    "마": 90,
+    "막": 367,
+    "만": 52,
+    "많": 158,
+    "맏": 1143,
+    "말": 177,
+    "맑": 1159,
+    "맘": 1251,
+    "맙": 1154,
+    "맛": 904,
+    "망": 304,
+    "맞": 505,
+    "맡": 906,
+    "매": 211,
+    "맥": 564,
+    "맨": 710,
+    "맵": 1010,
+    "맷": 798,
+    "맸": 1237,
+    "맹": 530,
+    "맺": 1027,
+    "머": 386,
+    "먹": 481,
+    "먼": 462,
+    "멀": 568,
+    "멈": 970,
+    "멋": 848,
+    "멍": 851,
+    "멎": 1294,
+    "메": 313,
+    "멘": 581,
+    "멜": 849,
+    "멤": 713,
+    "멧": 1072,
+    "며": 86,
+    "멱": 1277,
+    "면": 80,
+    "멸": 707,
+    "몄": 1275,
+    "명": 103,
+    "몇": 272,
+    "모": 111,
+    "목": 263,
+    "몫": 1125,
+    "몬": 685,
+    "몰": 532,
+    "몸": 570,
+    "몹": 1012,
+    "못": 396,
+    "몽": 621,
+    "묘": 765,
+    "무": 135,
+    "묵": 759,
+    "묶": 925,
+    "문": 73,
+    "묻": 746,
+    "물": 94,
+    "묽": 1291,
+    "뭄": 1313,
+    "뭅": 902,
+    "뭇": 826,
+    "뭉": 1269,
+    "뭍": 1200,
+    "뭐": 1030,
+    "뭔": 928,
+    "뮤": 608,
+    "뮬": 790,
+    "므": 493,
+    "미": 77,
+    "믹": 1061,
+    "민": 264,
+    "믿": 861,
+    "밀": 473,
+    "밈": 1266,
+    "밉": 1288,
+    "밋": 967,
+    "밍": 806,
+    "및": 245,
+    "밑": 646,
+    "바": 145,
+    "박": 331,
+    "밖": 703,
+    "반": 142,
+    "받": 171,
+    "발": 98,
+    "밝": 389,
+    "밟": 1132,
+    "밤": 446,
+    "밥": 1031,
+    "밧": 1224,
+    "방": 113,
+    "밭": 900,
+    "배": 269,
+    "백": 383,
+    "밴": 586,
+    "밸": 816,
+    "뱀": 818,
+    "뱃": 1199,
+    "뱅": 738,
+    "뱉": 1201,
+    "버": 226,
+    "벅": 1328,
+    "번": 207,
+    "벌": 535,
+    "범": 449,
+    "법": 160,
+    "벗": 588,
+    "벙": 1169,
+    "베": 278,
+    "벡": 875,
+    "벤": 606,
+    "벨": 476,
+    "벳": 1007,
+    "벼": 894,
+    "벽": 578,
+    "변": 262,
+    "별": 341,
+    "볍": 969,
+    "병": 365,
+    "보": 47,
+    "복": 268,
+    "본": 253,
+    "볼": 374,
+    "봄": 915,
+    "봅": 1024,
+    "봉": 619,
+    "봐": 697,
+    "뵈": 1198,
+    "뵐": 1225,
+    "뵙": 1145,
+    "부": 46,
+    "북": 289,
+    "분": 101,
+    "불": 172,
+    "붉": 1194,
+    "붐": 1050,
+    "붓": 1211,
+    "붕": 753,
+    "붙": 562,
+    "뷰": 786,
+    "뷴": 957,
+    "브": 257,
+    "븐": 749,
+    "블": 478,
+    "비": 72,
+    "빅": 784,
+    "빈": 802,
+    "빌": 684,
+    "빔": 1311,
+    "빕": 1094,
+    "빗": 886,
+    "빙": 542,
+    "빚": 1110,
+    "빛": 508,
+    "빠": 456,
+    "빡": 945,
+    "빤": 1264,
+    "빨": 660,
+    "빵": 1011,
+    "빼": 897,
+    "뺀": 1299,
+    "뺨": 1176,
+    "뻐": 1092,
+    "뻔": 944,
+    "뻗": 1001,
+    "뻤": 1263,
+    "뼈": 922,
+    "뽐": 1203,
+    "뽑": 893,
+    "뽕": 1246,
+    "뾰": 934,
+    "뿌": 696,
+    "뿐": 503,
+    "뿔": 1138,
+    "뿜": 1215,
+    "쁘": 1014,
+    "쁜": 825,
+    "쁨": 1099,
+    "쁩": 966,
+    "삐": 950,
+    "사": 21,
+    "삭": 701,
+    "삯": 1121,
+    "산": 206,
+    "살": 352,
+    "삶": 879,
+    "삼": 896,
+    "삽": 795,
+    "샀": 1183,
+    "상": 51,
+    "샅": 1184,
+    "새": 276,
+    "색": 523,
+    "샌": 940,
+    "샘": 1023,
+    "샛": 960,
+    "생": 116,
+    "샤": 648,
+    "샬": 1089,
+    "샵": 1049,
+    "서": 19,
+    "석": 283,
+    "섞": 979,
+    "선": 129,
+    "설": 287,
+    "섬": 404,
+    "섭": 800,
+    "섯": 652,
+    "섰": 903,
+    "성": 64,
+    "세": 84,
+    "섹": 1048,
+    "센": 437,
+    "셀": 480,
+    "셈": 688,
+    "셉": 1086,
+    "셋": 1113,
+    "셔": 651,
+    "션": 375,
+    "셨": 946,
+    "셰": 736,
+    "소": 74,
+    "속": 190,
+    "손": 330,
+    "솔": 850,
+    "솜": 1273,
+    "솟": 1074,
+    "송": 448,
+    "쇄": 597,
+    "쇠": 763,
+    "쇼": 406,
+    "수": 28,
+    "숙": 409,
+    "순": 345,
+    "술": 337,
+    "숨": 596,
+    "숫": 714,
+    "숭": 984,
+    "숯": 1270,
+    "숲": 1257,
+    "쉬": 656,
+    "쉰": 1286,
+    "쉴": 952,
+    "쉼": 1326,
+    "쉽": 401,
+    "슈": 512,
+    "슐": 856,
+    "슘": 1018,
+    "스": 26,
+    "슨": 483,
+    "슬": 418,
+    "슭": 1300,
+    "슴": 821,
+    "습": 24,
+    "슷": 438,
+    "승": 306,
+    "시": 35,
+    "식": 130,
+    "신": 107,
+    "싣": 846,
+    "실": 178,
+    "싫": 1213,
+    "심": 241,
+    "십": 400,
+    "싱": 545,
+    "싶": 751,
+    "싸": 560,
+    "싹": 1175,
+    "싼": 682,
+    "쌀": 912,
+    "쌈": 1222,
+    "쌉": 1051,
+    "쌌": 1216,
+    "쌍": 828,
+    "쌓": 741,
+    "써": 659,
+    "썩": 987,
+    "썰": 847,
+    "썼": 1117,
+    "썽": 1314,
+    "쏘": 916,
+    "쏟": 999,
+    "쏠": 1212,
+    "쑤": 1127,
+    "쑥": 1220,
+    "쓰": 384,
+    "쓴": 1129,
+    "쓸": 719,
+    "씀": 993,
+    "씁": 1108,
+    "씌": 1147,
+    "씨": 486,
+    "씩": 943,
+    "씬": 516,
+    "씻": 1114,
+    "아": 33,
+    "악": 344,
+    "안": 109,
+    "앉": 803,
+    "않": 117,
+    "알": 185,
+    "앎": 1204,
+    "앓": 1137,
+    "암": 428,
+    "압": 548,
+    "앗": 949,
+    "았": 212,
+    "앙": 574,
+    "앞": 440,
+    "애": 390,
+    "액": 589,
+    "앨": 770,
+    "앱": 954,
+    "앵": 1053,
+    "야": 102,
+    "약": 204,
+    "얀": 1226,
+    "얄": 1005,
+    "얇": 653,
+    "얌": 1249,
+    "얏": 878,
+    "양": 153,
+    "얕": 1235,
+    "얘": 1020,
+    "어": 31,
+    "억": 470,
+    "언": 218,
+    "얹": 1115,
+    "얻": 461,
+    "얼": 435,
+    "얽": 1158,
+    "엄": 566,
+    "업": 258,
+    "없": 183,
+    "엇": 1028,
+    "었": 61,
+    "엉": 1209,
+    "엎": 1106,
+    "에": 8,
+    "엔": 422,
+    "엘": 459,
+    "엠": 854,
+    "여": 63,
+    "역": 127,
+    "엮": 1223,
+    "연": 139,
+    "열": 244,
+    "염": 385,
+    "엽": 830,
+    "엾": 1207,
+    "엿": 1157,
+    "였": 236,
+    "영": 134,
+    "옆": 767,
+    "예": 192,
+    "옌": 1054,
+    "옛": 752,
+    "오": 85,
+    "옥": 840,
+    "온": 302,
+    "올": 216,
+    "옭": 1272,
+    "옮": 602,
+    "옳": 1093,
+    "옴": 1315,
+    "옵": 624,
+    "옷": 743,
+    "옹": 883,
+    "와": 70,
+    "완": 413,
+    "왈": 791,
+    "왔": 382,
+    "왕": 298,
+    "왜": 496,
+    "외": 323,
+    "왼": 814,
+    "요": 83,
+    "욕": 630,
+    "욥": 1131,
+    "용": 112,
+    "우": 75,
+    "욱": 772,
+    "운": 131,
+    "울": 319,
+    "움": 293,
+    "웁": 1165,
+    "웃": 647,
+    "웅": 1013,
+    "워": 324,
+    "원": 78,
+    "월": 188,
+    "웠": 951,
+    "웨": 392,
+    "웬": 1327,
+    "웰": 947,
+    "웹": 451,
+    "위": 54,
+    "윈": 700,
+    "윌": 1021,
+    "윗": 1096,
+    "윙": 961,
+    "유": 82,
+    "육": 407,
+    "윤": 754,
+    "율": 558,
+    "융": 579,
+    "으": 34,
+    "은": 13,
+    "을": 10,
+    "읊": 1260,
+    "음": 170,
+    "읍": 1075,
+    "응": 370,
+    "의": 9,
+    "이": 6,
+    "익": 538,
+    "인": 27,
+    "일": 38,
+    "읽": 571,
+    "잃": 702,
+    "임": 295,
+    "입": 76,
+    "잇": 708,
+    "있": 22,
+    "잉": 735,
+    "잊": 935,
+    "잎": 686,
+    "자": 37,
+    "작": 143,
+    "잔": 576,
+    "잖": 1285,
+    "잘": 410,
+    "잠": 416,
+    "잡": 398,
+    "잣": 1245,
+    "장": 62,
+    "잦": 958,
+    "재": 174,
+    "잭": 953,
+    "잿": 1241,
+    "쟁": 353,
+    "저": 209,
+    "적": 44,
+    "전": 43,
+    "절": 316,
+    "젊": 1082,
+    "점": 186,
+    "접": 271,
+    "젓": 1308,
+    "정": 57,
+    "젖": 932,
+    "제": 60,
+    "젝": 794,
+    "젠": 739,
+    "젤": 793,
+    "젯": 911,
+    "져": 256,
+    "젹": 965,
+    "젼": 832,
+    "졌": 267,
+    "조": 119,
+    "족": 252,
+    "존": 286,
+    "졸": 779,
+    "좀": 742,
+    "좁": 845,
+    "종": 159,
+    "좇": 1160,
+    "좋": 381,
+    "좌": 805,
+    "죄": 517,
+    "죔": 959,
+    "죠": 822,
+    "주": 55,
+    "죽": 599,
+    "준": 274,
+    "줄": 443,
+    "줌": 662,
+    "줍": 600,
+    "중": 92,
+    "줬": 889,
+    "쥐": 722,
+    "쥔": 974,
+    "즈": 279,
+    "즉": 485,
+    "즌": 583,
+    "즐": 490,
+    "즙": 1252,
+    "증": 250,
+    "지": 16,
+    "직": 234,
+    "진": 124,
+    "질": 179,
+    "짊": 1196,
+    "짐": 870,
+    "집": 284,
+    "짓": 711,
+    "징": 679,
+    "짖": 1104,
+    "짙": 1324,
+    "짚": 1231,
+    "짜": 694,
+    "짝": 799,
+    "짠": 1170,
+    "짤": 1278,
+    "짧": 605,
+    "째": 347,
+    "쨌": 1233,
+    "쨍": 817,
+    "쩌": 907,
+    "쩍": 1179,
+    "쩔": 1173,
+    "쩡": 1317,
+    "쪼": 1141,
+    "쪽": 233,
+    "쫒": 809,
+    "쫓": 1063,
+    "쬐": 1254,
+    "쭈": 1276,
+    "쭉": 1293,
+    "쭐": 1228,
+    "쯤": 874,
+    "찌": 1095,
+    "찍": 693,
+    "찐": 1289,
+    "찔": 1221,
+    "찢": 1150,
+    "차": 148,
+    "착": 342,
+    "찬": 546,
+    "찮": 775,
+    "찰": 326,
+    "참": 387,
+    "찼": 1174,
+    "창": 380,
+    "찾": 463,
+    "채": 395,
+    "책": 368,
+    "챈": 1271,
+    "챔": 612,
+    "챘": 1303,
+    "챙": 819,
+    "처": 220,
+    "척": 547,
+    "천": 328,
+    "철": 432,
+    "첨": 918,
+    "첩": 1304,
+    "첫": 423,
+    "청": 348,
+    "체": 146,
+    "쳐": 479,
+    "쳤": 631,
+    "초": 270,
+    "촉": 592,
+    "촌": 1248,
+    "촛": 1186,
+    "총": 497,
+    "촬": 811,
+    "최": 205,
+    "추": 243,
+    "축": 442,
+    "춘": 748,
+    "출": 217,
+    "춤": 1139,
+    "춥": 1083,
+    "춧": 1250,
+    "충": 373,
+    "춰": 771,
+    "췄": 924,
+    "취": 349,
+    "츠": 369,
+    "측": 444,
+    "츰": 1301,
+    "층": 553,
+    "치": 96,
+    "칙": 609,
+    "친": 403,
+    "칠": 519,
+    "침": 307,
+    "칩": 734,
+    "칫": 933,
+    "칭": 689,
+    "카": 187,
+    "칵": 1009,
+    "칸": 616,
+    "칼": 614,
+    "캄": 691,
+    "캐": 425,
+    "캔": 683,
+    "캘": 877,
+    "캠": 494,
+    "캡": 855,
+    "커": 394,
+    "컨": 584,
+    "컬": 865,
+    "컴": 556,
+    "컵": 1032,
+    "컷": 991,
+    "케": 288,
+    "켈": 1253,
+    "켓": 507,
+    "켜": 618,
+    "켰": 644,
+    "코": 231,
+    "콘": 663,
+    "콜": 649,
+    "콥": 1006,
+    "콧": 838,
+    "콩": 764,
+    "콰": 639,
+    "쾌": 1171,
+    "쾰": 841,
+    "쿄": 941,
+    "쿠": 427,
+    "쿡": 1026,
+    "쿨": 1077,
+    "쿼": 1008,
+    "퀴": 852,
+    "큄": 858,
+    "크": 141,
+    "큰": 358,
+    "클": 450,
+    "큼": 680,
+    "큽": 1244,
+    "키": 176,
+    "킨": 641,
+    "킬": 439,
+    "킴": 1205,
+    "킵": 962,
+    "킹": 475,
+    "타": 161,
+    "탁": 695,
+    "탄": 309,
+    "탈": 355,
+    "탉": 1217,
+    "탐": 501,
+    "탑": 676,
+    "탓": 1163,
+    "탔": 1229,
+    "탕": 632,
+    "태": 191,
+    "택": 397,
+    "탠": 796,
+    "탬": 1310,
+    "탱": 674,
+    "터": 132,
+    "턱": 1080,
+    "턴": 474,
+    "털": 665,
+    "텀": 873,
+    "텁": 1309,
+    "텅": 1234,
+    "테": 314,
+    "텍": 712,
+    "텐": 585,
+    "텔": 334,
+    "템": 534,
+    "텨": 1070,
+    "토": 237,
+    "톤": 733,
+    "톨": 657,
+    "톰": 998,
+    "톱": 522,
+    "통": 115,
+    "퇴": 723,
+    "투": 277,
+    "툰": 1298,
+    "툴": 1062,
+    "툼": 1044,
+    "퉁": 982,
+    "튜": 781,
+    "트": 68,
+    "특": 251,
+    "튼": 868,
+    "튿": 1118,
+    "틀": 603,
+    "틈": 866,
+    "티": 254,
+    "틱": 1067,
+    "틴": 655,
+    "팀": 363,
+    "팅": 1081,
+    "파": 166,
+    "팎": 1274,
+    "판": 308,
+    "팔": 468,
+    "팟": 981,
+    "팠": 1312,
+    "팡": 789,
+    "패": 343,
+    "팩": 1038,
+    "팬": 687,
+    "팸": 869,
+    "팽": 787,
+    "퍼": 469,
+    "펀": 914,
+    "페": 282,
+    "펜": 555,
+    "펠": 758,
+    "펴": 730,
+    "편": 305,
+    "펼": 737,
+    "폈": 1322,
+    "평": 364,
+    "폐": 539,
+    "포": 149,
+    "폭": 303,
+    "폰": 625,
+    "폴": 488,
+    "폼": 997,
+    "표": 200,
+    "푯": 1052,
+    "푸": 495,
+    "푼": 1162,
+    "풀": 511,
+    "품": 325,
+    "풍": 377,
+    "퓨": 543,
+    "프": 126,
+    "픈": 773,
+    "플": 362,
+    "픔": 1126,
+    "피": 213,
+    "픽": 379,
+    "핀": 778,
+    "필": 247,
+    "핌": 995,
+    "핍": 1035,
+    "핑": 467,
+    "하": 12,
+    "학": 156,
+    "한": 20,
+    "할": 69,
+    "핥": 1283,
+    "함": 167,
+    "합": 93,
+    "항": 189,
+    "해": 32,
+    "핵": 537,
+    "핼": 1037,
+    "햄": 913,
+    "햇": 777,
+    "했": 45,
+    "행": 100,
+    "향": 232,
+    "허": 335,
+    "헌": 776,
+    "헐": 1112,
+    "험": 225,
+    "헙": 1071,
+    "헛": 1107,
+    "헝": 797,
+    "헤": 429,
+    "헥": 836,
+    "헨": 890,
+    "헬": 813,
+    "헴": 1153,
+    "헷": 1036,
+    "혀": 414,
+    "혁": 595,
+    "현": 201,
+    "혈": 541,
+    "혐": 489,
+    "협": 453,
+    "혓": 926,
+    "혔": 520,
+    "형": 221,
+    "혜": 801,
+    "호": 175,
+    "혹": 447,
+    "혼": 492,
+    "홀": 552,
+    "홉": 1144,
+    "홍": 629,
+    "화": 81,
+    "확": 275,
+    "환": 322,
+    "활": 312,
+    "황": 424,
+    "홰": 1261,
+    "횃": 1218,
+    "회": 168,
+    "획": 417,
+    "횡": 706,
+    "효": 525,
+    "후": 138,
+    "훈": 675,
+    "훌": 727,
+    "훔": 909,
+    "훗": 1318,
+    "훨": 515,
+    "훼": 864,
+    "휘": 638,
+    "휠": 968,
+    "휩": 1148,
+    "휴": 531,
+    "휼": 1191,
+    "흉": 1161,
+    "흐": 477,
+    "흑": 572,
+    "흔": 434,
+    "흗": 1019,
+    "흘": 1069,
+    "흙": 1164,
+    "흠": 1111,
+    "흡": 762,
+    "흥": 650,
+    "흩": 717,
+    "희": 431,
+    "흰": 1022,
+    "히": 164,
+    "힌": 549,
+    "힐": 640,
+    "힘": 617
+  },
+  "kpq": {
+    "'": 33,
+    "-": 23,
+    "0": 32,
+    "1": 34,
+    "3": 35,
+    "4": 31,
+    "5": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 30,
+    "d": 16,
+    "e": 8,
+    "f": 26,
+    "g": 7,
+    "h": 25,
+    "i": 9,
+    "j": 27,
+    "k": 22,
+    "l": 12,
+    "m": 13,
+    "n": 6,
+    "o": 10,
+    "p": 24,
+    "r": 19,
+    "s": 15,
+    "t": 21,
+    "u": 14,
+    "w": 20,
+    "y": 18,
+    "z": 28,
+    "|": 4,
+    "́": 17,
+    "–": 29
+  },
+  "kps": {
+    "'": 30,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 29,
+    "d": 20,
+    "e": 8,
+    "f": 18,
+    "g": 23,
+    "h": 19,
+    "i": 6,
+    "j": 26,
+    "k": 11,
+    "l": 17,
+    "m": 12,
+    "n": 9,
+    "o": 7,
+    "p": 24,
+    "r": 16,
+    "s": 14,
+    "t": 10,
+    "u": 21,
+    "w": 13,
+    "y": 15,
+    "z": 28,
+    "|": 4,
+    "–": 27
+  },
+  "kpv": {
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "i": 23,
+    "|": 4,
+    "ö": 6,
+    "а": 9,
+    "б": 31,
+    "в": 16,
+    "г": 29,
+    "д": 10,
+    "е": 20,
+    "ж": 32,
+    "з": 25,
+    "и": 17,
+    "й": 21,
+    "к": 14,
+    "л": 13,
+    "м": 11,
+    "н": 8,
+    "о": 15,
+    "п": 26,
+    "р": 19,
+    "с": 5,
+    "т": 12,
+    "у": 22,
+    "ф": 38,
+    "х": 39,
+    "ц": 40,
+    "ч": 28,
+    "ш": 27,
+    "щ": 42,
+    "ъ": 33,
+    "ы": 7,
+    "ь": 18,
+    "э": 30,
+    "ю": 34,
+    "я": 24,
+    "ё": 35,
+    "і": 41,
+    "–": 37
+  },
+  "kpy": {
+    "'": 28,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 7,
+    "б": 37,
+    "в": 15,
+    "г": 9,
+    "д": 33,
+    "е": 22,
+    "ж": 40,
+    "з": 35,
+    "и": 12,
+    "й": 18,
+    "к": 13,
+    "л": 14,
+    "м": 17,
+    "н": 6,
+    "о": 16,
+    "п": 24,
+    "р": 31,
+    "с": 25,
+    "т": 8,
+    "у": 20,
+    "ф": 38,
+    "х": 39,
+    "ц": 36,
+    "ч": 19,
+    "ш": 41,
+    "ъ": 32,
+    "ы": 5,
+    "ь": 26,
+    "э": 10,
+    "ю": 30,
+    "я": 23,
+    "ё": 27,
+    "ӄ": 21,
+    "ӈ": 11,
+    "–": 34
+  },
+  "kpz": {
+    "'": 23,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 13,
+    "c": 16,
+    "e": 7,
+    "f": 25,
+    "i": 11,
+    "k": 6,
+    "l": 18,
+    "m": 14,
+    "n": 10,
+    "o": 5,
+    "p": 15,
+    "r": 19,
+    "s": 20,
+    "t": 8,
+    "u": 12,
+    "w": 17,
+    "y": 9,
+    "|": 4,
+    "ä": 24,
+    "ë": 26,
+    "ö": 22,
+    "ü": 27,
+    "ŋ": 21
+  },
+  "kqe": {
+    "'": 31,
+    "-": 23,
+    "0": 26,
+    "1": 28,
+    "2": 29,
+    "3": 32,
+    "4": 27,
+    "5": 30,
+    "6": 35,
+    "7": 36,
+    "8": 37,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "d": 15,
+    "e": 25,
+    "g": 7,
+    "h": 22,
+    "i": 9,
+    "j": 24,
+    "k": 11,
+    "l": 18,
+    "m": 12,
+    "n": 6,
+    "o": 8,
+    "p": 17,
+    "r": 21,
+    "s": 13,
+    "t": 14,
+    "u": 20,
+    "w": 16,
+    "y": 10,
+    "z": 34,
+    "|": 5
+  },
+  "kqp": {
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 32,
+    "d": 28,
+    "e": 6,
+    "g": 20,
+    "h": 22,
+    "i": 8,
+    "j": 27,
+    "k": 13,
+    "l": 19,
+    "m": 12,
+    "n": 7,
+    "o": 21,
+    "p": 29,
+    "r": 10,
+    "s": 24,
+    "t": 23,
+    "u": 25,
+    "w": 17,
+    "y": 15,
+    "|": 4,
+    "à": 33,
+    "á": 34,
+    "é": 16,
+    "ê": 18,
+    "ô": 9,
+    "ŋ": 26,
+    "ɓ": 31,
+    "ɗ": 30,
+    "ɲ": 14
+  },
+  "kqr": {
+    "'": 25,
+    "-": 23,
+    "0": 34,
+    "1": 36,
+    "2": 39,
+    "3": 41,
+    "4": 38,
+    "5": 43,
+    "6": 42,
+    "8": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 33,
+    "d": 11,
+    "e": 21,
+    "f": 30,
+    "g": 15,
+    "h": 26,
+    "i": 8,
+    "j": 24,
+    "k": 12,
+    "l": 18,
+    "m": 13,
+    "n": 7,
+    "o": 5,
+    "p": 19,
+    "r": 16,
+    "s": 14,
+    "t": 9,
+    "u": 10,
+    "v": 44,
+    "w": 20,
+    "y": 17,
+    "z": 32,
+    "|": 4,
+    "â": 29,
+    "ê": 37,
+    "î": 27,
+    "ô": 28,
+    "õ": 40,
+    "û": 31,
+    "–": 35
+  },
+  "kqy": {
+    "-": 171,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ሀ": 17,
+    "ሁ": 86,
+    "ሂ": 157,
+    "ሃ": 112,
+    "ሄ": 82,
+    "ህ": 13,
+    "ሆ": 125,
+    "ለ": 51,
+    "ሉ": 98,
+    "ሊ": 167,
+    "ላ": 85,
+    "ሌ": 73,
+    "ል": 59,
+    "ሎ": 57,
+    "መ": 34,
+    "ሙ": 90,
+    "ሚ": 116,
+    "ማ": 27,
+    "ሜ": 58,
+    "ም": 50,
+    "ሞ": 69,
+    "ሠ": 124,
+    "ሡ": 80,
+    "ሣ": 145,
+    "ሤ": 64,
+    "ሥ": 40,
+    "ሦ": 87,
+    "ረ": 18,
+    "ሩ": 72,
+    "ራ": 137,
+    "ሬ": 76,
+    "ር": 37,
+    "ሮ": 45,
+    "ሰ": 21,
+    "ሱ": 28,
+    "ሲ": 120,
+    "ሳ": 71,
+    "ሴ": 8,
+    "ስ": 11,
+    "ሶ": 42,
+    "ሸ": 52,
+    "ሹ": 104,
+    "ሺ": 172,
+    "ሻ": 136,
+    "ሼ": 89,
+    "ሽ": 63,
+    "ሾ": 117,
+    "ቀ": 134,
+    "ቁ": 84,
+    "ቂ": 166,
+    "ቃ": 150,
+    "ቄ": 55,
+    "ቅ": 100,
+    "ቆ": 109,
+    "በ": 41,
+    "ቡ": 83,
+    "ቢ": 162,
+    "ባ": 68,
+    "ቤ": 29,
+    "ብ": 49,
+    "ቦ": 78,
+    "ተ": 19,
+    "ቱ": 54,
+    "ታ": 61,
+    "ቴ": 30,
+    "ት": 33,
+    "ቶ": 20,
+    "ቸ": 56,
+    "ቹ": 147,
+    "ቻ": 156,
+    "ቼ": 133,
+    "ች": 111,
+    "ቾ": 81,
+    "ነ": 22,
+    "ኑ": 38,
+    "ኒ": 154,
+    "ና": 103,
+    "ኔ": 24,
+    "ን": 5,
+    "ኖ": 127,
+    "አ": 12,
+    "ኡ": 16,
+    "ኢ": 99,
+    "ኣ": 53,
+    "ኤ": 6,
+    "እ": 48,
+    "ኦ": 35,
+    "ከ": 43,
+    "ኩ": 75,
+    "ኪ": 161,
+    "ካ": 79,
+    "ኬ": 26,
+    "ክ": 62,
+    "ኮ": 10,
+    "ወ": 23,
+    "ዉ": 105,
+    "ዋ": 106,
+    "ዌ": 143,
+    "ው": 115,
+    "ዎ": 15,
+    "ዘ": 91,
+    "ዙ": 119,
+    "ዚ": 176,
+    "ዛ": 148,
+    "ዜ": 77,
+    "ዝ": 67,
+    "ዞ": 107,
+    "ዠ": 142,
+    "ዡ": 163,
+    "ዢ": 173,
+    "ዣ": 165,
+    "ዤ": 132,
+    "ዥ": 151,
+    "ዦ": 164,
+    "የ": 36,
+    "ዩ": 139,
+    "ዪ": 169,
+    "ያ": 96,
+    "ዬ": 9,
+    "ይ": 7,
+    "ዮ": 60,
+    "ደ": 44,
+    "ዱ": 97,
+    "ዲ": 131,
+    "ዳ": 129,
+    "ዴ": 25,
+    "ድ": 14,
+    "ዶ": 32,
+    "ዸ": 92,
+    "ዹ": 113,
+    "ዺ": 159,
+    "ዻ": 149,
+    "ዼ": 65,
+    "ዽ": 88,
+    "ዾ": 93,
+    "ጁ": 175,
+    "ጄ": 138,
+    "ጅ": 170,
+    "ገ": 31,
+    "ጉ": 66,
+    "ጊ": 177,
+    "ጋ": 135,
+    "ጌ": 39,
+    "ግ": 47,
+    "ጎ": 74,
+    "ጨ": 146,
+    "ጩ": 102,
+    "ጫ": 144,
+    "ጬ": 122,
+    "ጭ": 121,
+    "ጮ": 126,
+    "ጰ": 128,
+    "ጱ": 168,
+    "ጳ": 158,
+    "ጴ": 141,
+    "ጵ": 130,
+    "ጶ": 155,
+    "ጸ": 114,
+    "ጹ": 110,
+    "ጺ": 152,
+    "ጻ": 153,
+    "ጼ": 108,
+    "ጽ": 94,
+    "ጾ": 123,
+    "ፈ": 46,
+    "ፉ": 101,
+    "ፊ": 160,
+    "ፋ": 118,
+    "ፌ": 70,
+    "ፍ": 95,
+    "ፎ": 140,
+    "–": 174
+  },
+  "krc": {
+    "'": 38,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 18,
+    "в": 37,
+    "г": 13,
+    "д": 11,
+    "е": 9,
+    "ж": 30,
+    "з": 22,
+    "и": 8,
+    "й": 23,
+    "к": 14,
+    "л": 7,
+    "м": 20,
+    "н": 6,
+    "о": 24,
+    "п": 27,
+    "р": 12,
+    "с": 19,
+    "т": 17,
+    "у": 15,
+    "ф": 34,
+    "х": 25,
+    "ц": 39,
+    "ч": 29,
+    "ш": 28,
+    "ъ": 16,
+    "ы": 10,
+    "ь": 36,
+    "э": 26,
+    "ю": 21,
+    "я": 32,
+    "ё": 31,
+    "–": 33
+  },
+  "kri": {
+    "'": 32,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 21,
+    "c": 30,
+    "d": 8,
+    "e": 11,
+    "f": 22,
+    "g": 20,
+    "h": 29,
+    "i": 6,
+    "j": 26,
+    "k": 17,
+    "l": 15,
+    "m": 16,
+    "n": 5,
+    "o": 23,
+    "p": 19,
+    "r": 24,
+    "s": 13,
+    "t": 12,
+    "u": 18,
+    "v": 28,
+    "w": 14,
+    "y": 25,
+    "z": 27,
+    "|": 4,
+    "ɔ": 9,
+    "ɛ": 10
+  },
+  "krj": {
+    "'": 27,
+    "-": 24,
+    "0": 30,
+    "1": 34,
+    "2": 35,
+    "3": 38,
+    "4": 37,
+    "5": 39,
+    "6": 41,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 15,
+    "e": 23,
+    "f": 28,
+    "g": 7,
+    "h": 21,
+    "i": 8,
+    "j": 25,
+    "k": 9,
+    "l": 18,
+    "m": 12,
+    "n": 6,
+    "o": 10,
+    "p": 16,
+    "q": 32,
+    "r": 17,
+    "s": 13,
+    "t": 14,
+    "u": 11,
+    "v": 31,
+    "w": 22,
+    "x": 36,
+    "y": 19,
+    "z": 29,
+    "|": 5,
+    "—": 33
+  },
+  "krl": {
+    "'": 26,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 32,
+    "d": 30,
+    "e": 9,
+    "f": 27,
+    "g": 28,
+    "h": 15,
+    "i": 5,
+    "j": 22,
+    "k": 14,
+    "l": 10,
+    "m": 17,
+    "n": 8,
+    "o": 13,
+    "p": 20,
+    "r": 23,
+    "s": 18,
+    "t": 7,
+    "u": 12,
+    "v": 19,
+    "y": 21,
+    "z": 33,
+    "|": 4,
+    "ä": 11,
+    "ö": 24,
+    "č": 25,
+    "š": 16,
+    "–": 29
+  },
+  "krr": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ក": 17,
+    "ខ": 48,
+    "គ": 21,
+    "ឃ": 54,
+    "ង": 16,
+    "ច": 34,
+    "ឆ": 45,
+    "ជ": 52,
+    "ឈ": 55,
+    "ញ": 44,
+    "ដ": 29,
+    "ឋ": 49,
+    "ឌ": 35,
+    "ឍ": 56,
+    "ណ": 42,
+    "ត": 13,
+    "ថ": 57,
+    "ទ": 40,
+    "ធ": 59,
+    "ន": 15,
+    "ប": 14,
+    "ផ": 53,
+    "ព": 33,
+    "ភ": 58,
+    "ម": 7,
+    "យ": 20,
+    "រ": 8,
+    "ល": 39,
+    "វ": 26,
+    "ឝ": 43,
+    "ឞ": 47,
+    "ស": 22,
+    "ហ": 18,
+    "ឡ": 25,
+    "អ": 10,
+    "ា": 5,
+    "ិ": 27,
+    "ី": 28,
+    "ឹ": 30,
+    "ឺ": 38,
+    "ុ": 36,
+    "ូ": 23,
+    "ួ": 51,
+    "ើ": 31,
+    "ឿ": 46,
+    "ៀ": 41,
+    "ែ": 6,
+    "ៃ": 24,
+    "ោ": 50,
+    "ំ": 37,
+    "ះ": 32,
+    "៉": 9,
+    "៊": 12,
+    "់": 19,
+    "្": 11,
+    "៝": 60
+  },
+  "krs": {
+    "'": 51,
+    "-": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "b": 19,
+    "c": 49,
+    "d": 9,
+    "f": 46,
+    "g": 10,
+    "h": 23,
+    "j": 29,
+    "k": 8,
+    "l": 28,
+    "m": 16,
+    "n": 11,
+    "p": 27,
+    "r": 26,
+    "s": 15,
+    "t": 31,
+    "v": 42,
+    "w": 41,
+    "y": 21,
+    "z": 37,
+    "|": 4,
+    "á": 5,
+    "â": 39,
+    "ã": 7,
+    "ä": 12,
+    "é": 13,
+    "ê": 48,
+    "ë": 14,
+    "í": 6,
+    "î": 33,
+    "ï": 24,
+    "ó": 25,
+    "ô": 50,
+    "õ": 18,
+    "ö": 35,
+    "ú": 34,
+    "û": 52,
+    "ü": 40,
+    "ă": 38,
+    "ĕ": 45,
+    "ĩ": 17,
+    "ĭ": 44,
+    "ŋ": 32,
+    "ŏ": 43,
+    "ũ": 30,
+    "ŭ": 47,
+    "ḷ": 36,
+    "ẽ": 20,
+    "ꞌ": 22
+  },
+  "kru": {
+    "'": 28,
+    "-": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 45,
+    "ं": 27,
+    "अ": 21,
+    "आ": 23,
+    "इ": 42,
+    "ई": 39,
+    "उ": 40,
+    "ऊ": 54,
+    "ऎ": 44,
+    "ए": 36,
+    "ऐ": 63,
+    "ऒ": 61,
+    "ओ": 37,
+    "औ": 57,
+    "क": 12,
+    "ख": 25,
+    "ग": 16,
+    "घ": 60,
+    "च": 26,
+    "छ": 53,
+    "ज": 34,
+    "झ": 56,
+    "ट": 41,
+    "ठ": 49,
+    "ड": 33,
+    "ढ": 59,
+    "त": 15,
+    "थ": 48,
+    "द": 20,
+    "ध": 46,
+    "न": 5,
+    "प": 35,
+    "फ": 55,
+    "ब": 22,
+    "भ": 52,
+    "म": 11,
+    "य": 31,
+    "र": 7,
+    "ल": 18,
+    "व": 51,
+    "स": 9,
+    "ह": 17,
+    "़": 24,
+    "ा": 6,
+    "ि": 14,
+    "ी": 13,
+    "ु": 30,
+    "ू": 32,
+    "ृ": 62,
+    "ॆ": 43,
+    "े": 10,
+    "ै": 38,
+    "ॊ": 47,
+    "ो": 29,
+    "ौ": 50,
+    "्": 8,
+    "‍": 19
+  },
+  "ksb": {
+    "'": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 28,
+    "d": 24,
+    "e": 9,
+    "f": 26,
+    "g": 16,
+    "h": 17,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 13,
+    "m": 11,
+    "n": 7,
+    "o": 14,
+    "p": 27,
+    "r": 31,
+    "s": 20,
+    "t": 19,
+    "u": 8,
+    "v": 22,
+    "w": 10,
+    "y": 15,
+    "z": 18,
+    "|": 4,
+    "ú": 32,
+    "ṃ": 33,
+    "ṅ": 29,
+    "—": 30
+  },
+  "ksr": {
+    "'": 36,
+    "-": 28,
+    "0": 30,
+    "1": 31,
+    "2": 34,
+    "3": 38,
+    "4": 33,
+    "5": 35,
+    "6": 37,
+    "7": 32,
+    "8": 40,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 41,
+    "d": 24,
+    "e": 9,
+    "f": 29,
+    "g": 14,
+    "h": 26,
+    "i": 8,
+    "j": 15,
+    "k": 12,
+    "l": 16,
+    "m": 10,
+    "n": 11,
+    "o": 6,
+    "p": 25,
+    "q": 22,
+    "r": 19,
+    "s": 20,
+    "t": 17,
+    "u": 13,
+    "v": 42,
+    "w": 18,
+    "y": 23,
+    "z": 27,
+    "|": 5,
+    "ŋ": 7
+  },
+  "kss": {
+    "'": 40,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 31,
+    "c": 27,
+    "d": 16,
+    "e": 22,
+    "f": 35,
+    "g": 36,
+    "h": 20,
+    "i": 15,
+    "j": 39,
+    "k": 19,
+    "l": 9,
+    "m": 14,
+    "n": 8,
+    "o": 12,
+    "p": 29,
+    "s": 28,
+    "t": 30,
+    "u": 23,
+    "v": 37,
+    "w": 25,
+    "y": 21,
+    "|": 4,
+    "ā": 7,
+    "ē": 33,
+    "ī": 18,
+    "ŋ": 13,
+    "ō": 24,
+    "ū": 26,
+    "ɔ": 17,
+    "ɛ": 10,
+    "̄": 11,
+    "̍": 5,
+    "᷄": 32,
+    "᷅": 34
+  },
+  "ktb": {
+    "-": 135,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ሀ": 36,
+    "ሁ": 34,
+    "ሂ": 128,
+    "ሃ": 40,
+    "ሄ": 84,
+    "ህ": 21,
+    "ሆ": 43,
+    "ለ": 32,
+    "ሉ": 95,
+    "ሊ": 112,
+    "ላ": 73,
+    "ሌ": 70,
+    "ል": 39,
+    "ሎ": 91,
+    "ሏ": 172,
+    "መ": 8,
+    "ሙ": 74,
+    "ሚ": 102,
+    "ማ": 42,
+    "ሜ": 57,
+    "ም": 13,
+    "ሞ": 46,
+    "ሟ": 163,
+    "ረ": 31,
+    "ሩ": 60,
+    "ሪ": 93,
+    "ራ": 79,
+    "ሬ": 61,
+    "ር": 19,
+    "ሮ": 71,
+    "ሯ": 174,
+    "ሰ": 17,
+    "ሱ": 29,
+    "ሲ": 85,
+    "ሳ": 64,
+    "ሴ": 52,
+    "ስ": 10,
+    "ሶ": 54,
+    "ሷ": 176,
+    "ሸ": 111,
+    "ሹ": 151,
+    "ሺ": 156,
+    "ሻ": 137,
+    "ሼ": 140,
+    "ሽ": 86,
+    "ሾ": 113,
+    "ሿ": 178,
+    "ቀ": 37,
+    "ቁ": 132,
+    "ቂ": 154,
+    "ቃ": 103,
+    "ቄ": 96,
+    "ቅ": 87,
+    "ቆ": 78,
+    "ቋ": 184,
+    "በ": 16,
+    "ቡ": 83,
+    "ቢ": 114,
+    "ባ": 68,
+    "ቤ": 89,
+    "ብ": 63,
+    "ቦ": 90,
+    "ቧ": 165,
+    "ተ": 9,
+    "ቱ": 67,
+    "ቲ": 99,
+    "ታ": 53,
+    "ቴ": 55,
+    "ት": 12,
+    "ቶ": 38,
+    "ቷ": 170,
+    "ቸ": 138,
+    "ቹ": 49,
+    "ቺ": 119,
+    "ቻ": 161,
+    "ቼ": 133,
+    "ች": 22,
+    "ቾ": 100,
+    "ቿ": 160,
+    "ነ": 24,
+    "ኑ": 25,
+    "ኒ": 23,
+    "ና": 50,
+    "ኔ": 15,
+    "ን": 5,
+    "ኖ": 20,
+    "ኗ": 152,
+    "ኘ": 177,
+    "ኙ": 182,
+    "አ": 6,
+    "ኡ": 59,
+    "ኢ": 66,
+    "ኣ": 51,
+    "ኤ": 45,
+    "እ": 7,
+    "ኦ": 35,
+    "ከ": 18,
+    "ኩ": 27,
+    "ኪ": 106,
+    "ካ": 94,
+    "ኬ": 44,
+    "ክ": 33,
+    "ኮ": 48,
+    "ኳ": 185,
+    "ወ": 115,
+    "ዉ": 108,
+    "ዊ": 136,
+    "ዋ": 92,
+    "ዌ": 168,
+    "ው": 116,
+    "ዎ": 56,
+    "ዕ": 14,
+    "ዘ": 65,
+    "ዙ": 158,
+    "ዚ": 166,
+    "ዛ": 139,
+    "ዜ": 130,
+    "ዝ": 144,
+    "ዞ": 155,
+    "ዟ": 180,
+    "ዦ": 181,
+    "የ": 75,
+    "ዩ": 157,
+    "ዪ": 147,
+    "ያ": 97,
+    "ዬ": 28,
+    "ይ": 62,
+    "ዮ": 41,
+    "ደ": 26,
+    "ዱ": 105,
+    "ዲ": 110,
+    "ዳ": 47,
+    "ዴ": 142,
+    "ድ": 77,
+    "ዶ": 72,
+    "ዷ": 179,
+    "ጀ": 117,
+    "ጁ": 164,
+    "ጂ": 169,
+    "ጃ": 88,
+    "ጄ": 131,
+    "ጅ": 104,
+    "ጆ": 143,
+    "ገ": 11,
+    "ጉ": 69,
+    "ጊ": 129,
+    "ጋ": 82,
+    "ጌ": 122,
+    "ግ": 76,
+    "ጎ": 98,
+    "ጠ": 30,
+    "ጡ": 58,
+    "ጢ": 120,
+    "ጣ": 109,
+    "ጤ": 127,
+    "ጥ": 107,
+    "ጦ": 121,
+    "ጧ": 173,
+    "ጨ": 118,
+    "ጩ": 125,
+    "ጪ": 159,
+    "ጫ": 146,
+    "ጬ": 153,
+    "ጭ": 126,
+    "ጮ": 123,
+    "ጯ": 186,
+    "ጰ": 149,
+    "ጱ": 175,
+    "ጲ": 171,
+    "ጳ": 141,
+    "ጴ": 150,
+    "ጵ": 134,
+    "ጶ": 167,
+    "ጷ": 183,
+    "ፈ": 80,
+    "ፉ": 81,
+    "ፊ": 162,
+    "ፋ": 148,
+    "ፌ": 145,
+    "ፍ": 101,
+    "ፎ": 124,
+    "ፏ": 187
+  },
+  "ktj": {
+    "'": 5,
+    "0": 36,
+    "2": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 15,
+    "c": 32,
+    "d": 22,
+    "e": 23,
+    "f": 34,
+    "g": 30,
+    "h": 25,
+    "i": 14,
+    "j": 33,
+    "k": 16,
+    "l": 12,
+    "m": 18,
+    "n": 8,
+    "o": 28,
+    "p": 27,
+    "r": 29,
+    "s": 26,
+    "t": 21,
+    "u": 19,
+    "w": 24,
+    "y": 13,
+    "|": 4,
+    "ŋ": 35,
+    "ɔ": 17,
+    "ɛ": 11,
+    "ɩ": 6,
+    "ʋ": 7,
+    "ԑ": 31,
+    "‐": 9,
+    "꞊": 20
+  },
+  "kub": {
+    "'": 39,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 33,
+    "d": 27,
+    "e": 10,
+    "f": 29,
+    "g": 23,
+    "h": 36,
+    "i": 9,
+    "j": 32,
+    "k": 11,
+    "l": 38,
+    "m": 14,
+    "n": 5,
+    "o": 35,
+    "p": 30,
+    "r": 18,
+    "s": 15,
+    "t": 7,
+    "u": 12,
+    "v": 43,
+    "w": 13,
+    "x": 37,
+    "y": 20,
+    "z": 40,
+    "|": 4,
+    "à": 42,
+    "á": 19,
+    "è": 46,
+    "é": 25,
+    "ì": 45,
+    "í": 24,
+    "ó": 34,
+    "ù": 41,
+    "ú": 22,
+    "ā": 17,
+    "ē": 26,
+    "ī": 8,
+    "ō": 31,
+    "ū": 21,
+    "ḿ": 44,
+    "–": 47
+  },
+  "kue": {
+    "'": 32,
+    "-": 23,
+    "0": 27,
+    "1": 29,
+    "2": 30,
+    "3": 35,
+    "4": 33,
+    "5": 34,
+    "6": 36,
+    "7": 31,
+    "8": 37,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 16,
+    "e": 6,
+    "f": 28,
+    "g": 9,
+    "h": 24,
+    "i": 8,
+    "j": 25,
+    "k": 12,
+    "l": 14,
+    "m": 11,
+    "n": 7,
+    "o": 10,
+    "p": 20,
+    "r": 13,
+    "s": 22,
+    "t": 21,
+    "u": 15,
+    "v": 26,
+    "w": 18,
+    "y": 17,
+    "|": 4
+  },
+  "kum": {
+    "0": 37,
+    "1": 38,
+    "2": 39,
+    "4": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 16,
+    "в": 28,
+    "г": 9,
+    "д": 14,
+    "е": 8,
+    "ж": 33,
+    "з": 22,
+    "и": 10,
+    "й": 25,
+    "к": 17,
+    "л": 7,
+    "м": 19,
+    "н": 6,
+    "о": 21,
+    "п": 26,
+    "р": 11,
+    "с": 20,
+    "т": 18,
+    "у": 15,
+    "ф": 36,
+    "х": 34,
+    "ц": 41,
+    "ч": 29,
+    "ш": 27,
+    "ъ": 13,
+    "ы": 12,
+    "ь": 24,
+    "э": 31,
+    "ю": 23,
+    "я": 30,
+    "ё": 32,
+    "–": 35
+  },
+  "kus": {
+    "'": 16,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "d": 18,
+    "e": 8,
+    "f": 26,
+    "g": 20,
+    "h": 29,
+    "i": 6,
+    "j": 27,
+    "k": 13,
+    "l": 12,
+    "m": 10,
+    "n": 7,
+    "o": 11,
+    "p": 21,
+    "r": 23,
+    "s": 14,
+    "t": 19,
+    "u": 9,
+    "v": 28,
+    "w": 24,
+    "y": 17,
+    "z": 25,
+    "|": 4,
+    "ŋ": 22
+  },
+  "kvn": {
+    "'": 47,
+    "-": 12,
+    "0": 44,
+    "1": 36,
+    "2": 37,
+    "3": 40,
+    "4": 39,
+    "5": 41,
+    "6": 42,
+    "7": 48,
+    "8": 43,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 22,
+    "d": 18,
+    "e": 6,
+    "f": 34,
+    "g": 11,
+    "h": 20,
+    "i": 9,
+    "j": 25,
+    "k": 17,
+    "l": 7,
+    "m": 10,
+    "n": 8,
+    "o": 16,
+    "p": 15,
+    "q": 38,
+    "r": 24,
+    "s": 21,
+    "t": 14,
+    "u": 13,
+    "v": 33,
+    "w": 23,
+    "x": 46,
+    "y": 27,
+    "z": 26,
+    "|": 5,
+    "á": 32,
+    "é": 29,
+    "í": 30,
+    "ó": 35,
+    "ú": 28,
+    "—": 31
+  },
+  "kvw": {
+    "'": 25,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 28,
+    "d": 18,
+    "e": 9,
+    "f": 26,
+    "g": 6,
+    "h": 24,
+    "i": 8,
+    "j": 27,
+    "k": 17,
+    "l": 10,
+    "m": 13,
+    "n": 7,
+    "o": 11,
+    "p": 16,
+    "r": 19,
+    "s": 14,
+    "t": 15,
+    "u": 12,
+    "w": 21,
+    "y": 22,
+    "|": 4
+  },
+  "kwd": {
+    "'": 6,
+    "0": 26,
+    "1": 28,
+    "2": 29,
+    "4": 30,
+    "6": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "d": 21,
+    "e": 13,
+    "f": 16,
+    "g": 11,
+    "h": 25,
+    "i": 8,
+    "j": 23,
+    "k": 18,
+    "l": 12,
+    "m": 14,
+    "n": 7,
+    "o": 9,
+    "p": 24,
+    "r": 17,
+    "s": 19,
+    "t": 15,
+    "u": 10,
+    "v": 27,
+    "w": 20,
+    "|": 5,
+    "—": 32
+  },
+  "kwf": {
+    "'": 8,
+    "0": 29,
+    "1": 32,
+    "2": 34,
+    "4": 31,
+    "5": 33,
+    "6": 30,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 26,
+    "d": 20,
+    "e": 9,
+    "f": 17,
+    "g": 18,
+    "h": 24,
+    "i": 6,
+    "j": 23,
+    "k": 10,
+    "l": 15,
+    "m": 14,
+    "n": 7,
+    "o": 12,
+    "p": 25,
+    "r": 16,
+    "s": 13,
+    "t": 19,
+    "u": 11,
+    "v": 27,
+    "w": 21,
+    "y": 28,
+    "|": 5
+  },
+  "kwi": {
+    "'": 32,
+    "0": 41,
+    "1": 43,
+    "2": 44,
+    "4": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 22,
+    "d": 23,
+    "e": 17,
+    "f": 36,
+    "g": 33,
+    "h": 19,
+    "i": 7,
+    "j": 24,
+    "k": 9,
+    "l": 20,
+    "m": 13,
+    "n": 6,
+    "o": 21,
+    "p": 12,
+    "q": 40,
+    "r": 15,
+    "s": 10,
+    "t": 11,
+    "u": 8,
+    "v": 38,
+    "w": 16,
+    "x": 42,
+    "y": 26,
+    "z": 14,
+    "|": 5,
+    "á": 34,
+    "ã": 39,
+    "é": 31,
+    "í": 29,
+    "ñ": 25,
+    "ó": 37,
+    "ú": 27,
+    "ĩ": 35,
+    "ɨ": 18,
+    "—": 30
+  },
+  "kxc": {
+    "-": 139,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ሀ": 38,
+    "ሁ": 88,
+    "ሂ": 133,
+    "ሃ": 75,
+    "ሄ": 72,
+    "ህ": 104,
+    "ሆ": 110,
+    "ለ": 20,
+    "ሉ": 105,
+    "ሊ": 134,
+    "ላ": 77,
+    "ሌ": 45,
+    "ል": 33,
+    "ሎ": 56,
+    "መ": 10,
+    "ሙ": 87,
+    "ሚ": 112,
+    "ማ": 64,
+    "ሜ": 59,
+    "ም": 35,
+    "ሞ": 57,
+    "ረ": 12,
+    "ሩ": 109,
+    "ሪ": 131,
+    "ራ": 111,
+    "ሬ": 58,
+    "ር": 24,
+    "ሮ": 62,
+    "ሰ": 16,
+    "ሱ": 52,
+    "ሲ": 121,
+    "ሳ": 100,
+    "ሴ": 14,
+    "ስ": 23,
+    "ሶ": 61,
+    "ሸ": 40,
+    "ሹ": 120,
+    "ሺ": 124,
+    "ሻ": 127,
+    "ሼ": 53,
+    "ሽ": 39,
+    "ሾ": 37,
+    "ቀ": 18,
+    "ቁ": 51,
+    "ቂ": 137,
+    "ቃ": 90,
+    "ቄ": 76,
+    "ቅ": 80,
+    "ቆ": 60,
+    "ተ": 7,
+    "ቱ": 70,
+    "ቲ": 107,
+    "ታ": 55,
+    "ቴ": 31,
+    "ት": 30,
+    "ቶ": 28,
+    "ቸ": 102,
+    "ቹ": 128,
+    "ቺ": 117,
+    "ቻ": 94,
+    "ች": 113,
+    "ቾ": 95,
+    "ነ": 13,
+    "ኑ": 97,
+    "ኒ": 119,
+    "ና": 67,
+    "ኔ": 22,
+    "ን": 5,
+    "ኖ": 34,
+    "ኘ": 115,
+    "ኙ": 135,
+    "ኛ": 138,
+    "ኜ": 114,
+    "ኝ": 122,
+    "ኞ": 142,
+    "አ": 8,
+    "ኡ": 50,
+    "ኢ": 136,
+    "ኣ": 73,
+    "ኤ": 26,
+    "እ": 9,
+    "ኦ": 15,
+    "ከ": 6,
+    "ኩ": 48,
+    "ኪ": 71,
+    "ካ": 65,
+    "ኬ": 47,
+    "ክ": 21,
+    "ኮ": 32,
+    "ኸ": 49,
+    "ኹ": 108,
+    "ኺ": 145,
+    "ኻ": 79,
+    "ኼ": 103,
+    "ኽ": 123,
+    "ኾ": 78,
+    "ወ": 46,
+    "ዊ": 146,
+    "ዋ": 44,
+    "ዌ": 82,
+    "ው": 43,
+    "ዎ": 118,
+    "የ": 42,
+    "ዩ": 130,
+    "ያ": 85,
+    "ዬ": 27,
+    "ይ": 17,
+    "ዮ": 69,
+    "ደ": 25,
+    "ዱ": 91,
+    "ዲ": 86,
+    "ዳ": 66,
+    "ዴ": 19,
+    "ድ": 36,
+    "ዶ": 84,
+    "ጨ": 74,
+    "ጩ": 99,
+    "ጪ": 129,
+    "ጫ": 96,
+    "ጬ": 106,
+    "ጭ": 116,
+    "ጮ": 132,
+    "ጰ": 126,
+    "ጳ": 141,
+    "ጴ": 140,
+    "ጵ": 125,
+    "ጶ": 143,
+    "ፈ": 63,
+    "ፉ": 93,
+    "ፊ": 144,
+    "ፋ": 98,
+    "ፌ": 92,
+    "ፍ": 101,
+    "ፎ": 54,
+    "ፐ": 11,
+    "ፑ": 83,
+    "ፒ": 68,
+    "ፓ": 89,
+    "ፔ": 81,
+    "ፕ": 29,
+    "ፖ": 41
+  },
+  "kxf": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 25,
+    "c": 34,
+    "d": 21,
+    "e": 26,
+    "g": 33,
+    "h": 6,
+    "i": 35,
+    "j": 38,
+    "k": 8,
+    "l": 17,
+    "m": 24,
+    "n": 19,
+    "o": 30,
+    "p": 15,
+    "r": 37,
+    "s": 13,
+    "t": 10,
+    "u": 28,
+    "w": 29,
+    "y": 27,
+    "|": 4,
+    "è": 22,
+    "î": 16,
+    "ò": 18,
+    "ô": 12,
+    "û": 11,
+    "ě": 23,
+    "ǎ": 14,
+    "ǐ": 31,
+    "ǒ": 20,
+    "ǔ": 32,
+    "̂": 39,
+    "̌": 5,
+    "̤": 9,
+    "ṳ": 36
+  },
+  "kxm": {
+    "'": 46,
+    "-": 49,
+    "0": 43,
+    "1": 44,
+    "2": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 47,
+    "|": 4,
+    "ก": 18,
+    "ค": 29,
+    "ง": 13,
+    "จ": 25,
+    "ช": 40,
+    "ซ": 28,
+    "ญ": 37,
+    "ด": 16,
+    "ต": 26,
+    "ท": 32,
+    "น": 7,
+    "บ": 21,
+    "ป": 20,
+    "พ": 33,
+    "ฟ": 42,
+    "ม": 17,
+    "ย": 9,
+    "ร": 12,
+    "ล": 23,
+    "ว": 27,
+    "อ": 5,
+    "ฮ": 15,
+    "ะ": 34,
+    "ั": 22,
+    "า": 14,
+    "ิ": 19,
+    "ี": 11,
+    "ึ": 39,
+    "ื": 30,
+    "ุ": 38,
+    "ู": 36,
+    "ฺ": 8,
+    "เ": 6,
+    "แ": 24,
+    "โ": 31,
+    "ไ": 41,
+    "็": 10,
+    "๋": 48,
+    "ํ": 35
+  },
+  "kxv": {
+    "'": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଁ": 19,
+    "ଂ": 41,
+    "ଃ": 40,
+    "ଅ": 36,
+    "ଆ": 24,
+    "ଇ": 23,
+    "ଈ": 39,
+    "ଉ": 43,
+    "ଊ": 42,
+    "ଏ": 13,
+    "କ": 17,
+    "ଗ": 31,
+    "ଙ": 37,
+    "ଚ": 28,
+    "ଜ": 20,
+    "ଞ": 35,
+    "ଟ": 33,
+    "ଡ": 32,
+    "ଣ": 27,
+    "ତ": 11,
+    "ଦ": 30,
+    "ନ": 12,
+    "ପ": 25,
+    "ବ": 26,
+    "ମ": 15,
+    "ର": 9,
+    "ଲ": 22,
+    "ସ": 14,
+    "ହ": 18,
+    "଼": 7,
+    "ା": 5,
+    "ି": 6,
+    "ୀ": 29,
+    "ୁ": 16,
+    "ୂ": 38,
+    "େ": 8,
+    "୍": 10,
+    "ୟ": 34,
+    "ୱ": 21
+  },
+  "kyb": {
+    "'": 18,
+    "-": 32,
+    "0": 31,
+    "1": 37,
+    "3": 36,
+    "4": 33,
+    "6": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 17,
+    "d": 23,
+    "e": 24,
+    "f": 27,
+    "g": 11,
+    "h": 10,
+    "i": 9,
+    "j": 21,
+    "k": 19,
+    "l": 22,
+    "m": 12,
+    "n": 6,
+    "o": 15,
+    "p": 20,
+    "q": 39,
+    "r": 25,
+    "s": 8,
+    "t": 16,
+    "u": 7,
+    "v": 29,
+    "w": 14,
+    "x": 30,
+    "y": 13,
+    "z": 28,
+    "|": 5,
+    "–": 35,
+    "—": 38
+  },
+  "kyc": {
+    "'": 27,
+    "-": 25,
+    "0": 26,
+    "1": 30,
+    "2": 32,
+    "3": 35,
+    "4": 29,
+    "6": 33,
+    "7": 34,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 39,
+    "a": 4,
+    "b": 20,
+    "c": 31,
+    "d": 16,
+    "e": 8,
+    "f": 38,
+    "g": 18,
+    "h": 28,
+    "i": 9,
+    "j": 23,
+    "k": 11,
+    "l": 15,
+    "m": 12,
+    "n": 7,
+    "o": 6,
+    "p": 10,
+    "r": 17,
+    "s": 19,
+    "t": 22,
+    "u": 14,
+    "v": 36,
+    "w": 21,
+    "y": 13,
+    "|": 5,
+    "ŋ": 24
+  },
+  "kyf": {
+    "'": 10,
+    "-": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 33,
+    "d": 23,
+    "e": 15,
+    "f": 31,
+    "g": 20,
+    "h": 26,
+    "i": 18,
+    "j": 32,
+    "k": 21,
+    "l": 8,
+    "m": 17,
+    "n": 7,
+    "o": 29,
+    "p": 28,
+    "s": 25,
+    "t": 27,
+    "u": 22,
+    "v": 34,
+    "w": 14,
+    "y": 11,
+    "z": 24,
+    "|": 4,
+    "ŋ": 30,
+    "ɔ": 13,
+    "ɛ": 19,
+    "ɩ": 6,
+    "ʋ": 12
+  },
+  "kyg": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "e": 6,
+    "f": 20,
+    "g": 12,
+    "h": 15,
+    "i": 9,
+    "k": 19,
+    "l": 16,
+    "m": 11,
+    "n": 10,
+    "o": 7,
+    "p": 18,
+    "s": 22,
+    "t": 17,
+    "u": 13,
+    "v": 21,
+    "y": 14,
+    "|": 5
+  },
+  "kyo": {
+    "'": 26,
+    "-": 27,
+    "3": 29,
+    "7": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 31,
+    "d": 18,
+    "e": 10,
+    "g": 7,
+    "h": 12,
+    "i": 8,
+    "j": 28,
+    "k": 15,
+    "l": 14,
+    "m": 16,
+    "n": 6,
+    "o": 9,
+    "p": 25,
+    "r": 21,
+    "s": 22,
+    "t": 20,
+    "u": 13,
+    "w": 24,
+    "y": 17,
+    "|": 4,
+    "é": 11,
+    "ó": 23
+  },
+  "kyq": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 30,
+    "d": 23,
+    "e": 6,
+    "g": 15,
+    "i": 10,
+    "j": 20,
+    "k": 8,
+    "l": 22,
+    "m": 13,
+    "n": 7,
+    "o": 16,
+    "p": 29,
+    "r": 19,
+    "s": 12,
+    "t": 11,
+    "u": 17,
+    "w": 33,
+    "y": 27,
+    "z": 32,
+    "|": 4,
+    "ŋ": 26,
+    "ƴ": 34,
+    "ɓ": 28,
+    "ɔ": 9,
+    "ɗ": 18,
+    "ɛ": 14,
+    "̰": 21,
+    "‐": 31
+  },
+  "kyu": {
+    "'": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "m": 40,
+    "|": 5,
+    "ꤊ": 20,
+    "ꤋ": 28,
+    "ꤌ": 24,
+    "ꤍ": 38,
+    "ꤎ": 39,
+    "ꤏ": 33,
+    "ꤐ": 37,
+    "ꤑ": 36,
+    "ꤒ": 13,
+    "ꤓ": 31,
+    "ꤔ": 14,
+    "ꤕ": 19,
+    "ꤖ": 32,
+    "ꤗ": 23,
+    "ꤘ": 26,
+    "ꤙ": 30,
+    "ꤚ": 16,
+    "ꤛ": 10,
+    "ꤜ": 22,
+    "ꤝ": 34,
+    "ꤞ": 25,
+    "ꤟ": 7,
+    "ꤠ": 35,
+    "ꤡ": 27,
+    "ꤢ": 4,
+    "ꤣ": 12,
+    "ꤤ": 17,
+    "ꤥ": 21,
+    "ꤦ": 29,
+    "ꤧ": 11,
+    "ꤨ": 18,
+    "ꤩ": 9,
+    "ꤪ": 15,
+    "꤬": 6,
+    "꤭": 8
+  },
+  "kyz": {
+    "'": 7,
+    "0": 37,
+    "4": 39,
+    "6": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 34,
+    "d": 28,
+    "e": 6,
+    "f": 25,
+    "g": 12,
+    "h": 35,
+    "i": 13,
+    "j": 10,
+    "k": 19,
+    "l": 29,
+    "m": 11,
+    "n": 17,
+    "o": 15,
+    "p": 14,
+    "r": 8,
+    "s": 24,
+    "t": 18,
+    "u": 9,
+    "v": 33,
+    "w": 16,
+    "x": 40,
+    "y": 20,
+    "z": 38,
+    "|": 4,
+    "ã": 22,
+    "õ": 31,
+    "ĩ": 27,
+    "ũ": 21,
+    "ẽ": 23,
+    "ỹ": 30,
+    "—": 26
+  },
+  "kzf": {
+    "'": 20,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 29,
+    "d": 21,
+    "e": 11,
+    "f": 27,
+    "g": 17,
+    "h": 26,
+    "i": 7,
+    "j": 23,
+    "k": 15,
+    "l": 18,
+    "m": 12,
+    "n": 6,
+    "o": 9,
+    "p": 13,
+    "r": 14,
+    "s": 16,
+    "t": 10,
+    "u": 8,
+    "w": 22,
+    "y": 25,
+    "z": 28,
+    "|": 5
+  },
+  "lac": {
+    "'": 9,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 6,
+    "d": 26,
+    "e": 14,
+    "f": 31,
+    "g": 32,
+    "h": 20,
+    "i": 8,
+    "j": 10,
+    "k": 37,
+    "l": 28,
+    "m": 21,
+    "n": 13,
+    "o": 15,
+    "p": 25,
+    "q": 22,
+    "r": 11,
+    "s": 17,
+    "t": 12,
+    "u": 7,
+    "v": 34,
+    "w": 24,
+    "x": 23,
+    "y": 19,
+    "z": 33,
+    "|": 4,
+    "á": 36,
+    "é": 30,
+    "í": 29,
+    "ñ": 38,
+    "ó": 35,
+    "ú": 27,
+    "ʌ": 18
+  },
+  "laj": {
+    "'": 26,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 19,
+    "c": 18,
+    "d": 17,
+    "e": 7,
+    "f": 27,
+    "g": 23,
+    "h": 30,
+    "i": 8,
+    "j": 24,
+    "k": 10,
+    "l": 21,
+    "m": 11,
+    "n": 9,
+    "o": 5,
+    "p": 22,
+    "r": 13,
+    "s": 29,
+    "t": 12,
+    "u": 15,
+    "v": 28,
+    "w": 14,
+    "y": 16,
+    "|": 4,
+    "ŋ": 20
+  },
+  "lam": {
+    "'": 26,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 22,
+    "d": 24,
+    "e": 8,
+    "f": 20,
+    "g": 25,
+    "i": 6,
+    "j": 27,
+    "k": 10,
+    "l": 9,
+    "m": 12,
+    "n": 11,
+    "o": 14,
+    "p": 19,
+    "s": 15,
+    "t": 17,
+    "u": 7,
+    "v": 30,
+    "w": 16,
+    "y": 18,
+    "|": 5,
+    "ŋ": 23,
+    "ŵ": 13,
+    "ʼ": 31,
+    "—": 29
+  },
+  "lao": {
+    "!": 116,
+    "%": 99,
+    "&": 112,
+    "'": 98,
+    "+": 117,
+    ",": 92,
+    "-": 90,
+    ".": 89,
+    "/": 97,
+    "0": 60,
+    "1": 67,
+    "2": 74,
+    "3": 86,
+    "4": 82,
+    "5": 81,
+    "6": 87,
+    "7": 91,
+    "8": 84,
+    "9": 80,
+    ":": 94,
+    ";": 100,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 118,
+    "[": 109,
+    "]": 110,
+    "a": 40,
+    "b": 72,
+    "c": 62,
+    "d": 64,
+    "e": 47,
+    "f": 79,
+    "g": 65,
+    "h": 59,
+    "i": 53,
+    "j": 83,
+    "k": 73,
+    "l": 57,
+    "m": 66,
+    "n": 48,
+    "o": 52,
+    "p": 71,
+    "q": 95,
+    "r": 50,
+    "s": 56,
+    "t": 58,
+    "u": 63,
+    "v": 85,
+    "w": 76,
+    "x": 93,
+    "y": 75,
+    "z": 88,
+    "¥": 101,
+    "á": 103,
+    "ã": 120,
+    "ç": 106,
+    "é": 113,
+    "í": 105,
+    "ó": 122,
+    "ú": 107,
+    "ü": 104,
+    "ກ": 7,
+    "ຂ": 26,
+    "ຄ": 32,
+    "ງ": 8,
+    "ຈ": 33,
+    "ຊ": 36,
+    "ຍ": 23,
+    "ດ": 9,
+    "ຕ": 29,
+    "ຖ": 44,
+    "ທ": 18,
+    "ນ": 5,
+    "ບ": 21,
+    "ປ": 25,
+    "ຜ": 55,
+    "ຝ": 78,
+    "ພ": 35,
+    "ຟ": 70,
+    "ມ": 17,
+    "ຢ": 46,
+    "ຣ": 69,
+    "ລ": 20,
+    "ວ": 16,
+    "ສ": 19,
+    "ຫ": 28,
+    "ອ": 13,
+    "ຮ": 37,
+    "ະ": 14,
+    "ັ": 11,
+    "າ": 4,
+    "ຳ": 42,
+    "ິ": 27,
+    "ີ": 15,
+    "ຶ": 54,
+    "ື": 31,
+    "ຸ": 41,
+    "ູ": 38,
+    "ົ": 22,
+    "ຼ": 43,
+    "ຽ": 49,
+    "ເ": 12,
+    "ແ": 24,
+    "ໂ": 45,
+    "ໃ": 30,
+    "ໄ": 34,
+    "ໆ": 68,
+    "່": 6,
+    "້": 10,
+    "໊": 102,
+    "໋": 121,
+    "໌": 77,
+    "ໍ": 39,
+    "ໜ": 51,
+    "ໝ": 61,
+    "​": 111,
+    "–": 119,
+    "—": 108,
+    "‘": 114,
+    "’": 115,
+    "”": 96
+  },
+  "las": {
+    "'": 32,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "c": 27,
+    "d": 34,
+    "e": 6,
+    "f": 30,
+    "h": 29,
+    "i": 13,
+    "k": 14,
+    "l": 18,
+    "m": 17,
+    "n": 7,
+    "o": 24,
+    "p": 22,
+    "r": 11,
+    "s": 9,
+    "t": 8,
+    "u": 25,
+    "w": 16,
+    "y": 23,
+    "|": 4,
+    "ñ": 26,
+    "đ": 33,
+    "ŋ": 31,
+    "ɔ": 19,
+    "ɖ": 28,
+    "ɛ": 10,
+    "ɩ": 12,
+    "ʋ": 15,
+    "ꞌ": 20
+  },
+  "lat": {
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 20,
+    "c": 15,
+    "d": 16,
+    "e": 5,
+    "f": 24,
+    "g": 22,
+    "h": 23,
+    "i": 6,
+    "l": 17,
+    "m": 12,
+    "n": 11,
+    "o": 14,
+    "p": 18,
+    "q": 21,
+    "r": 13,
+    "s": 8,
+    "t": 7,
+    "u": 9,
+    "v": 19,
+    "x": 25,
+    "y": 26,
+    "z": 27,
+    "|": 4,
+    "ì": 31,
+    "í": 30,
+    "—": 28
+  },
+  "lav": {
+    "!": 70,
+    "\"": 69,
+    "$": 74,
+    "%": 57,
+    "&": 73,
+    "'": 56,
+    "+": 63,
+    ",": 55,
+    "-": 49,
+    ".": 37,
+    "/": 52,
+    "0": 34,
+    "1": 35,
+    "2": 41,
+    "3": 45,
+    "4": 46,
+    "5": 42,
+    "6": 44,
+    "7": 48,
+    "8": 47,
+    "9": 43,
+    ":": 51,
+    ";": 62,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 27,
+    "d": 18,
+    "e": 9,
+    "f": 32,
+    "g": 26,
+    "h": 38,
+    "i": 6,
+    "j": 21,
+    "k": 14,
+    "l": 17,
+    "m": 16,
+    "n": 12,
+    "o": 15,
+    "p": 19,
+    "q": 60,
+    "r": 10,
+    "s": 7,
+    "t": 8,
+    "u": 11,
+    "v": 20,
+    "w": 54,
+    "x": 53,
+    "y": 50,
+    "z": 22,
+    "|": 4,
+    "£": 68,
+    "¨": 59,
+    "°": 65,
+    "²": 67,
+    "ç": 72,
+    "õ": 64,
+    "ā": 13,
+    "č": 40,
+    "ē": 24,
+    "ģ": 36,
+    "ī": 23,
+    "ķ": 39,
+    "ļ": 31,
+    "ņ": 30,
+    "š": 28,
+    "ū": 29,
+    "ž": 33,
+    "​": 71,
+    "–": 61,
+    "—": 66,
+    "”": 58
+  },
+  "law": {
+    "'": 11,
+    "1": 27,
+    "5": 32,
+    "8": 28,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 30,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 21,
+    "e": 7,
+    "g": 15,
+    "h": 25,
+    "i": 9,
+    "j": 19,
+    "k": 22,
+    "l": 13,
+    "m": 12,
+    "n": 8,
+    "o": 6,
+    "p": 16,
+    "r": 23,
+    "s": 17,
+    "t": 14,
+    "u": 10,
+    "v": 29,
+    "w": 24,
+    "y": 20,
+    "|": 5
+  },
+  "lbj": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "་": 4,
+    "ཀ": 22,
+    "ཁ": 21,
+    "ག": 12,
+    "ང": 10,
+    "ཅ": 28,
+    "ཆ": 31,
+    "ཇ": 47,
+    "ཉ": 38,
+    "ཏ": 25,
+    "ཐ": 41,
+    "ད": 11,
+    "ན": 7,
+    "པ": 24,
+    "ཕ": 33,
+    "བ": 16,
+    "མ": 14,
+    "ཙ": 43,
+    "ཚ": 32,
+    "ཛ": 39,
+    "ཝ": 59,
+    "ཞ": 27,
+    "ཟ": 30,
+    "འ": 17,
+    "ཡ": 20,
+    "ར": 19,
+    "ལ": 18,
+    "ཤ": 26,
+    "ས": 6,
+    "ཧ": 42,
+    "ཨ": 44,
+    "ཱ": 36,
+    "ི": 8,
+    "ུ": 15,
+    "ེ": 13,
+    "ོ": 9,
+    "ྐ": 35,
+    "ྒ": 40,
+    "ྔ": 46,
+    "ྕ": 51,
+    "ྗ": 55,
+    "ྙ": 50,
+    "ྟ": 37,
+    "ྡ": 45,
+    "ྣ": 54,
+    "ྤ": 48,
+    "ྦ": 57,
+    "ྨ": 52,
+    "ྩ": 49,
+    "ྫ": 53,
+    "ྭ": 56,
+    "ྱ": 23,
+    "ྲ": 29,
+    "ླ": 34,
+    "ྷ": 58
+  },
+  "lbw": {
+    "'": 16,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 18,
+    "c": 27,
+    "d": 22,
+    "e": 9,
+    "g": 14,
+    "h": 21,
+    "i": 7,
+    "j": 26,
+    "k": 13,
+    "l": 17,
+    "m": 11,
+    "n": 8,
+    "o": 4,
+    "p": 20,
+    "r": 15,
+    "s": 19,
+    "t": 12,
+    "u": 10,
+    "w": 23,
+    "y": 25,
+    "|": 5
+  },
+  "lcp": {
+    "'": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "k": 44,
+    "|": 4,
+    "ก": 17,
+    "ค": 34,
+    "ฆ": 36,
+    "ง": 20,
+    "จ": 24,
+    "ช": 41,
+    "ซ": 18,
+    "ฌ": 42,
+    "ญ": 33,
+    "ด": 35,
+    "ต": 11,
+    "ท": 39,
+    "น": 21,
+    "บ": 37,
+    "ป": 10,
+    "พ": 31,
+    "ฟ": 40,
+    "ม": 9,
+    "ย": 13,
+    "ร": 26,
+    "ล": 22,
+    "ว": 16,
+    "อ": 5,
+    "ฮ": 7,
+    "ะ": 8,
+    "ั": 15,
+    "า": 19,
+    "ิ": 32,
+    "ี": 27,
+    "ึ": 28,
+    "ื": 14,
+    "ุ": 23,
+    "ู": 38,
+    "เ": 6,
+    "แ": 30,
+    "โ": 12,
+    "ไ": 25,
+    "่": 29
+  },
+  "lee": {
+    "'": 28,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 17,
+    "b": 11,
+    "c": 26,
+    "d": 18,
+    "e": 36,
+    "f": 56,
+    "g": 31,
+    "h": 24,
+    "i": 19,
+    "j": 41,
+    "k": 25,
+    "l": 14,
+    "m": 13,
+    "n": 6,
+    "o": 57,
+    "p": 42,
+    "r": 16,
+    "s": 23,
+    "t": 27,
+    "u": 39,
+    "v": 52,
+    "w": 12,
+    "y": 9,
+    "z": 21,
+    "|": 4,
+    "à": 22,
+    "á": 20,
+    "â": 53,
+    "ã": 50,
+    "è": 46,
+    "é": 32,
+    "ê": 49,
+    "ì": 30,
+    "í": 29,
+    "ò": 43,
+    "ó": 37,
+    "ô": 62,
+    "õ": 58,
+    "ù": 38,
+    "ú": 44,
+    "ě": 60,
+    "ĩ": 54,
+    "ń": 33,
+    "ŋ": 55,
+    "ũ": 61,
+    "ǎ": 48,
+    "ǐ": 51,
+    "ǒ": 63,
+    "ǔ": 59,
+    "ǹ": 34,
+    "ɔ": 15,
+    "ə": 5,
+    "ɛ": 10,
+    "̀": 8,
+    "́": 7,
+    "̂": 66,
+    "̃": 35,
+    "̌": 45,
+    "ṍ": 65,
+    "ṹ": 64,
+    "ẽ": 47
+  },
+  "lef": {
+    "'": 33,
+    "-": 22,
+    "2": 39,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "d": 17,
+    "e": 13,
+    "g": 26,
+    "h": 34,
+    "i": 6,
+    "k": 10,
+    "l": 12,
+    "m": 14,
+    "n": 7,
+    "o": 20,
+    "p": 25,
+    "r": 30,
+    "s": 19,
+    "t": 16,
+    "u": 9,
+    "v": 31,
+    "w": 21,
+    "y": 18,
+    "|": 4,
+    "á": 27,
+    "é": 29,
+    "í": 32,
+    "ó": 35,
+    "ú": 28,
+    "ń": 36,
+    "ƒ": 23,
+    "ɔ": 8,
+    "ɛ": 15,
+    "́": 24,
+    "ḿ": 37
+  },
+  "lem": {
+    "'": 36,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 25,
+    "d": 30,
+    "e": 12,
+    "f": 31,
+    "g": 33,
+    "h": 26,
+    "i": 15,
+    "j": 32,
+    "k": 17,
+    "l": 22,
+    "m": 13,
+    "n": 6,
+    "o": 27,
+    "s": 23,
+    "t": 19,
+    "u": 21,
+    "w": 28,
+    "y": 14,
+    "|": 4,
+    "á": 10,
+    "é": 18,
+    "í": 16,
+    "ó": 29,
+    "ú": 24,
+    "ŋ": 20,
+    "ɔ": 7,
+    "ɛ": 11,
+    "ʼ": 34,
+    "́": 8
+  },
+  "lew": {
+    "'": 32,
+    "-": 24,
+    "1": 33,
+    "2": 36,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 30,
+    "d": 20,
+    "e": 12,
+    "f": 26,
+    "g": 19,
+    "h": 23,
+    "i": 7,
+    "j": 25,
+    "k": 16,
+    "l": 17,
+    "m": 10,
+    "n": 6,
+    "o": 9,
+    "p": 13,
+    "r": 14,
+    "s": 15,
+    "t": 11,
+    "u": 8,
+    "v": 22,
+    "w": 31,
+    "y": 21,
+    "z": 29,
+    "|": 5,
+    "˻": 27,
+    "˼": 28,
+    "—": 34
+  },
+  "lex": {
+    "'": 18,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 28,
+    "d": 16,
+    "e": 8,
+    "f": 26,
+    "g": 24,
+    "h": 17,
+    "i": 6,
+    "j": 27,
+    "k": 15,
+    "l": 10,
+    "m": 11,
+    "n": 7,
+    "o": 12,
+    "p": 19,
+    "r": 9,
+    "s": 23,
+    "t": 13,
+    "u": 14,
+    "w": 20,
+    "y": 21,
+    "z": 29,
+    "|": 5
+  },
+  "lgg": {
+    "'": 23,
+    "-": 31,
+    "0": 32,
+    "1": 33,
+    "3": 35,
+    "5": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "c": 26,
+    "d": 13,
+    "e": 7,
+    "f": 25,
+    "g": 22,
+    "h": 29,
+    "i": 5,
+    "j": 27,
+    "k": 16,
+    "l": 15,
+    "m": 12,
+    "n": 10,
+    "o": 11,
+    "p": 21,
+    "q": 36,
+    "r": 8,
+    "s": 20,
+    "t": 19,
+    "u": 9,
+    "v": 24,
+    "w": 28,
+    "y": 14,
+    "z": 18,
+    "|": 4,
+    "­": 30
+  },
+  "lgl": {
+    "'": 31,
+    "-": 30,
+    "0": 29,
+    "1": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 26,
+    "d": 21,
+    "e": 9,
+    "f": 13,
+    "g": 17,
+    "h": 24,
+    "i": 6,
+    "j": 23,
+    "k": 14,
+    "l": 7,
+    "m": 12,
+    "n": 19,
+    "o": 8,
+    "p": 25,
+    "r": 16,
+    "s": 15,
+    "t": 18,
+    "u": 10,
+    "v": 27,
+    "w": 20,
+    "y": 28,
+    "|": 5,
+    "ꞌ": 11
+  },
+  "lhu": {
+    "'": 16,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 29,
+    "c": 22,
+    "d": 28,
+    "e": 7,
+    "f": 33,
+    "g": 23,
+    "h": 6,
+    "i": 12,
+    "j": 31,
+    "k": 15,
+    "l": 17,
+    "m": 24,
+    "n": 19,
+    "o": 14,
+    "p": 21,
+    "r": 32,
+    "s": 26,
+    "t": 10,
+    "u": 11,
+    "v": 13,
+    "w": 9,
+    "y": 18,
+    "z": 34,
+    "|": 4,
+    "ˆ": 25,
+    "ˉ": 27,
+    "ˍ": 20,
+    "ˬ": 8
+  },
+  "lia": {
+    "'": 30,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "d": 20,
+    "e": 12,
+    "f": 27,
+    "g": 26,
+    "h": 17,
+    "i": 6,
+    "k": 9,
+    "l": 23,
+    "m": 14,
+    "n": 7,
+    "o": 19,
+    "p": 25,
+    "r": 24,
+    "s": 22,
+    "t": 15,
+    "u": 18,
+    "w": 21,
+    "y": 13,
+    "|": 4,
+    "ŋ": 8,
+    "ɔ": 16,
+    "ɛ": 11,
+    "—": 29
+  },
+  "lid": {
+    "'": 36,
+    "-": 22,
+    "0": 32,
+    "1": 26,
+    "2": 27,
+    "3": 29,
+    "4": 28,
+    "5": 30,
+    "6": 31,
+    "7": 35,
+    "8": 34,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 14,
+    "e": 7,
+    "g": 25,
+    "h": 11,
+    "i": 6,
+    "j": 23,
+    "k": 13,
+    "l": 17,
+    "m": 15,
+    "n": 8,
+    "o": 9,
+    "p": 24,
+    "r": 12,
+    "s": 19,
+    "t": 16,
+    "u": 10,
+    "w": 21,
+    "y": 20,
+    "|": 4,
+    "–": 37
+  },
+  "lif": {
+    "'": 48,
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "­": 50,
+    "ः": 19,
+    "अ": 43,
+    "आ": 28,
+    "इ": 32,
+    "उ": 44,
+    "ए": 41,
+    "ओ": 37,
+    "क": 10,
+    "ख": 27,
+    "ग": 29,
+    "घ": 42,
+    "ङ": 13,
+    "च": 35,
+    "छ": 38,
+    "ज": 36,
+    "त": 22,
+    "थ": 34,
+    "द": 31,
+    "ध": 40,
+    "न": 11,
+    "प": 24,
+    "फ": 33,
+    "ब": 23,
+    "भ": 39,
+    "म": 12,
+    "य": 26,
+    "र": 21,
+    "ल": 16,
+    "व": 30,
+    "स": 18,
+    "ह": 25,
+    "़": 8,
+    "ा": 6,
+    "ि": 14,
+    "ु": 15,
+    "े": 7,
+    "ै": 47,
+    "ो": 20,
+    "ौ": 49,
+    "्": 4,
+    "६": 51,
+    "ॽ": 17,
+    "‍": 9,
+    "–": 45,
+    "—": 52
+  },
+  "lin": {
+    "!": 52,
+    "\"": 56,
+    "$": 58,
+    "%": 57,
+    "&": 60,
+    "'": 48,
+    "+": 66,
+    ",": 47,
+    "-": 31,
+    ".": 30,
+    "/": 49,
+    "0": 27,
+    "1": 29,
+    "2": 32,
+    "3": 39,
+    "4": 38,
+    "5": 37,
+    "6": 40,
+    "7": 42,
+    "8": 41,
+    "9": 35,
+    ":": 45,
+    ";": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 51,
+    "[": 70,
+    "]": 71,
+    "a": 5,
+    "b": 12,
+    "c": 24,
+    "d": 21,
+    "e": 8,
+    "f": 26,
+    "g": 17,
+    "h": 25,
+    "i": 9,
+    "j": 34,
+    "k": 10,
+    "l": 13,
+    "m": 11,
+    "n": 7,
+    "o": 6,
+    "p": 18,
+    "q": 36,
+    "r": 22,
+    "s": 14,
+    "t": 16,
+    "u": 19,
+    "v": 28,
+    "w": 23,
+    "x": 43,
+    "y": 15,
+    "z": 20,
+    "|": 4,
+    "£": 73,
+    "¥": 64,
+    "°": 67,
+    "²": 74,
+    "à": 76,
+    "á": 62,
+    "â": 68,
+    "ã": 72,
+    "ç": 65,
+    "è": 46,
+    "é": 33,
+    "í": 59,
+    "î": 77,
+    "ï": 50,
+    "ó": 79,
+    "õ": 63,
+    "ö": 80,
+    "ú": 69,
+    "ü": 54,
+    "ł": 75,
+    "œ": 78,
+    "–": 53,
+    "—": 55,
+    "”": 61
+  },
+  "lip": {
+    "'": 45,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 12,
+    "d": 22,
+    "e": 9,
+    "f": 19,
+    "g": 35,
+    "h": 33,
+    "i": 7,
+    "k": 10,
+    "l": 17,
+    "m": 18,
+    "n": 8,
+    "o": 11,
+    "p": 24,
+    "r": 30,
+    "s": 13,
+    "t": 14,
+    "u": 15,
+    "v": 36,
+    "w": 23,
+    "y": 16,
+    "z": 40,
+    "|": 4,
+    "à": 43,
+    "á": 25,
+    "ã": 34,
+    "é": 26,
+    "í": 27,
+    "ó": 28,
+    "ú": 29,
+    "ĩ": 41,
+    "ń": 32,
+    "ũ": 39,
+    "ǝ": 5,
+    "ɔ": 21,
+    "ɛ": 20,
+    "̀": 46,
+    "́": 31,
+    "̃": 37,
+    "ḿ": 42,
+    "ẽ": 44
+  },
+  "lis": {
+    "'": 52,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "b": 51,
+    "g": 54,
+    "s": 53,
+    "w": 55,
+    "x": 57,
+    "z": 56,
+    "|": 4,
+    "ˍ": 17,
+    "ꓐ": 18,
+    "ꓑ": 48,
+    "ꓒ": 37,
+    "ꓓ": 32,
+    "ꓔ": 19,
+    "ꓕ": 27,
+    "ꓖ": 30,
+    "ꓗ": 20,
+    "ꓘ": 41,
+    "ꓙ": 25,
+    "ꓚ": 45,
+    "ꓛ": 40,
+    "ꓜ": 47,
+    "ꓝ": 42,
+    "ꓞ": 39,
+    "ꓟ": 9,
+    "ꓠ": 10,
+    "ꓡ": 11,
+    "ꓢ": 16,
+    "ꓣ": 38,
+    "ꓤ": 44,
+    "ꓥ": 28,
+    "ꓦ": 43,
+    "ꓧ": 49,
+    "ꓩ": 50,
+    "ꓪ": 13,
+    "ꓫ": 33,
+    "ꓬ": 6,
+    "ꓮ": 35,
+    "ꓯ": 14,
+    "ꓰ": 15,
+    "ꓱ": 21,
+    "ꓲ": 5,
+    "ꓳ": 8,
+    "ꓴ": 12,
+    "ꓵ": 29,
+    "ꓶ": 24,
+    "ꓷ": 36,
+    "ꓸ": 23,
+    "ꓹ": 22,
+    "ꓺ": 46,
+    "ꓻ": 34,
+    "ꓼ": 31,
+    "ꓽ": 7
+  },
+  "lit": {
+    "%": 59,
+    "'": 52,
+    "+": 60,
+    ",": 54,
+    "-": 51,
+    ".": 41,
+    "/": 55,
+    "0": 35,
+    "1": 38,
+    "2": 39,
+    "3": 45,
+    "4": 44,
+    "5": 43,
+    "6": 47,
+    "7": 49,
+    "8": 46,
+    "9": 42,
+    ":": 53,
+    ";": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 67,
+    "[": 69,
+    "]": 70,
+    "a": 6,
+    "b": 24,
+    "c": 32,
+    "d": 18,
+    "e": 10,
+    "f": 34,
+    "g": 21,
+    "h": 36,
+    "i": 5,
+    "j": 20,
+    "k": 14,
+    "l": 16,
+    "m": 15,
+    "n": 12,
+    "o": 8,
+    "p": 17,
+    "q": 56,
+    "r": 11,
+    "s": 7,
+    "t": 9,
+    "u": 13,
+    "v": 19,
+    "w": 50,
+    "x": 48,
+    "y": 23,
+    "z": 33,
+    "|": 4,
+    "°": 61,
+    "²": 66,
+    "ç": 63,
+    "é": 68,
+    "õ": 62,
+    "ö": 65,
+    "ú": 64,
+    "ü": 57,
+    "ą": 28,
+    "č": 31,
+    "ė": 22,
+    "ę": 37,
+    "į": 29,
+    "š": 26,
+    "ū": 30,
+    "ų": 25,
+    "ž": 27,
+    "–": 40
+  },
+  "lje": {
+    "'": 27,
+    "-": 24,
+    "0": 29,
+    "1": 34,
+    "2": 32,
+    "3": 36,
+    "4": 31,
+    "5": 37,
+    "6": 33,
+    "7": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 30,
+    "d": 21,
+    "e": 10,
+    "f": 28,
+    "g": 20,
+    "h": 13,
+    "i": 7,
+    "j": 26,
+    "k": 15,
+    "l": 16,
+    "m": 14,
+    "n": 9,
+    "o": 6,
+    "p": 17,
+    "r": 18,
+    "s": 23,
+    "t": 12,
+    "u": 11,
+    "w": 19,
+    "y": 25,
+    "|": 4,
+    "ꞌ": 8
+  },
+  "ljp": {
+    "'": 29,
+    "-": 24,
+    "0": 30,
+    "1": 34,
+    "2": 35,
+    "3": 36,
+    "4": 32,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 26,
+    "d": 17,
+    "e": 10,
+    "f": 27,
+    "g": 16,
+    "h": 18,
+    "i": 6,
+    "j": 19,
+    "k": 8,
+    "l": 12,
+    "m": 11,
+    "n": 7,
+    "o": 20,
+    "p": 22,
+    "r": 15,
+    "s": 13,
+    "t": 14,
+    "u": 9,
+    "v": 31,
+    "w": 25,
+    "y": 23,
+    "z": 28,
+    "|": 5,
+    "ë": 37
+  },
+  "llg": {
+    "'": 21,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 27,
+    "d": 16,
+    "e": 6,
+    "f": 20,
+    "g": 19,
+    "h": 17,
+    "i": 9,
+    "k": 14,
+    "l": 8,
+    "m": 12,
+    "n": 7,
+    "o": 10,
+    "p": 24,
+    "r": 25,
+    "s": 13,
+    "t": 15,
+    "u": 11,
+    "w": 26,
+    "x": 28,
+    "y": 22,
+    "|": 5
+  },
+  "lln": {
+    "'": 35,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 38,
+    "a": 5,
+    "b": 17,
+    "c": 29,
+    "d": 15,
+    "e": 6,
+    "g": 8,
+    "h": 32,
+    "i": 9,
+    "j": 21,
+    "k": 11,
+    "l": 16,
+    "m": 18,
+    "n": 7,
+    "o": 10,
+    "p": 27,
+    "r": 19,
+    "s": 20,
+    "t": 22,
+    "u": 12,
+    "w": 26,
+    "y": 13,
+    "|": 4,
+    "á": 24,
+    "é": 28,
+    "í": 33,
+    "ó": 30,
+    "ú": 34,
+    "ŋ": 14,
+    "ɓ": 31,
+    "ɗ": 25,
+    "̃": 37,
+    "ỹ": 23
+  },
+  "lme": {
+    "'": 18,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 26,
+    "d": 14,
+    "e": 17,
+    "f": 27,
+    "g": 25,
+    "h": 23,
+    "i": 9,
+    "j": 34,
+    "k": 11,
+    "l": 30,
+    "m": 8,
+    "n": 7,
+    "o": 20,
+    "p": 35,
+    "r": 13,
+    "s": 16,
+    "t": 21,
+    "u": 10,
+    "v": 31,
+    "w": 12,
+    "y": 15,
+    "z": 22,
+    "|": 4,
+    "â": 28,
+    "ê": 38,
+    "î": 36,
+    "ô": 33,
+    "û": 37,
+    "ŋ": 29,
+    "ɓ": 24,
+    "ɗ": 32,
+    "ə": 6
+  },
+  "lnd": {
+    "'": 26,
+    "-": 25,
+    "0": 29,
+    "1": 31,
+    "2": 32,
+    "4": 33,
+    "5": 38,
+    "6": 34,
+    "7": 35,
+    "8": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 24,
+    "d": 13,
+    "e": 8,
+    "g": 15,
+    "h": 14,
+    "i": 9,
+    "j": 28,
+    "k": 10,
+    "l": 11,
+    "m": 12,
+    "n": 6,
+    "o": 22,
+    "p": 19,
+    "r": 17,
+    "s": 20,
+    "t": 16,
+    "u": 7,
+    "v": 37,
+    "w": 27,
+    "y": 21,
+    "|": 4,
+    "é": 23,
+    "—": 30
+  },
+  "lns": {
+    "-": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 31,
+    "c": 44,
+    "d": 22,
+    "e": 8,
+    "f": 26,
+    "g": 39,
+    "h": 37,
+    "i": 7,
+    "j": 33,
+    "k": 11,
+    "l": 34,
+    "m": 16,
+    "n": 5,
+    "o": 13,
+    "p": 43,
+    "r": 17,
+    "s": 15,
+    "t": 32,
+    "u": 23,
+    "v": 14,
+    "w": 10,
+    "y": 6,
+    "z": 29,
+    "|": 4,
+    "à": 30,
+    "á": 21,
+    "è": 24,
+    "é": 28,
+    "ì": 20,
+    "í": 18,
+    "ò": 36,
+    "ó": 38,
+    "ù": 19,
+    "ú": 41,
+    "ě": 45,
+    "ŋ": 35,
+    "ǎ": 47,
+    "ǐ": 49,
+    "ǒ": 46,
+    "ǔ": 48,
+    "ə": 12,
+    "ʼ": 25,
+    "̀": 27,
+    "́": 40,
+    "̌": 50
+  },
+  "lob": {
+    "'": 35,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 34,
+    "d": 13,
+    "e": 22,
+    "f": 30,
+    "g": 16,
+    "h": 12,
+    "i": 20,
+    "j": 29,
+    "k": 11,
+    "l": 18,
+    "m": 25,
+    "n": 7,
+    "o": 19,
+    "p": 24,
+    "r": 9,
+    "s": 21,
+    "t": 14,
+    "u": 27,
+    "v": 43,
+    "w": 28,
+    "y": 33,
+    "|": 4,
+    "ã": 31,
+    "õ": 40,
+    "ĩ": 42,
+    "ũ": 38,
+    "ƴ": 37,
+    "ǝ": 44,
+    "ɓ": 41,
+    "ɔ": 10,
+    "ə": 36,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ʋ": 17,
+    "̃": 23,
+    "ι": 26,
+    "ẽ": 39
+  },
+  "lok": {
+    "'": 29,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 28,
+    "a": 5,
+    "b": 14,
+    "c": 27,
+    "d": 21,
+    "e": 12,
+    "f": 25,
+    "g": 9,
+    "h": 15,
+    "i": 7,
+    "j": 26,
+    "k": 17,
+    "l": 10,
+    "m": 20,
+    "n": 6,
+    "o": 18,
+    "p": 23,
+    "q": 30,
+    "s": 24,
+    "t": 19,
+    "u": 13,
+    "w": 16,
+    "y": 22,
+    "|": 4,
+    "ɔ": 11,
+    "ɛ": 8
+  },
+  "lom": {
+    "'": 54,
+    "-": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 30,
+    "c": 57,
+    "d": 25,
+    "e": 9,
+    "f": 31,
+    "g": 12,
+    "h": 46,
+    "i": 6,
+    "j": 44,
+    "k": 21,
+    "l": 8,
+    "m": 19,
+    "n": 10,
+    "o": 15,
+    "p": 27,
+    "s": 24,
+    "t": 13,
+    "u": 11,
+    "v": 20,
+    "w": 16,
+    "y": 26,
+    "z": 17,
+    "|": 4,
+    "à": 35,
+    "á": 34,
+    "ã": 42,
+    "è": 33,
+    "é": 23,
+    "ê": 58,
+    "ì": 43,
+    "í": 41,
+    "ò": 51,
+    "ó": 50,
+    "õ": 40,
+    "ù": 47,
+    "ú": 45,
+    "ĩ": 49,
+    "ŋ": 32,
+    "ũ": 38,
+    "ƃ": 28,
+    "ɓ": 56,
+    "ɔ": 14,
+    "ɛ": 7,
+    "ɣ": 18,
+    "ʋ": 29,
+    "̀": 36,
+    "́": 37,
+    "̃": 39,
+    "ẁ": 52,
+    "ẃ": 55,
+    "ẽ": 48,
+    "‐": 22,
+    "—": 53
+  },
+  "lon": {
+    "'": 27,
+    "0": 30,
+    "1": 31,
+    "2": 32,
+    "3": 34,
+    "4": 33,
+    "5": 36,
+    "6": 37,
+    "7": 38,
+    "8": 39,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 24,
+    "d": 18,
+    "e": 7,
+    "f": 28,
+    "g": 25,
+    "h": 14,
+    "i": 8,
+    "j": 29,
+    "k": 15,
+    "l": 6,
+    "m": 12,
+    "n": 10,
+    "o": 9,
+    "p": 21,
+    "r": 19,
+    "s": 22,
+    "t": 16,
+    "u": 11,
+    "v": 20,
+    "w": 13,
+    "y": 17,
+    "z": 26,
+    "|": 5
+  },
+  "loq": {
+    "'": 29,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "d": 26,
+    "e": 9,
+    "f": 28,
+    "g": 22,
+    "h": 24,
+    "i": 13,
+    "k": 12,
+    "l": 14,
+    "m": 8,
+    "n": 7,
+    "o": 6,
+    "p": 23,
+    "r": 37,
+    "s": 17,
+    "t": 19,
+    "u": 21,
+    "v": 27,
+    "w": 18,
+    "y": 10,
+    "z": 25,
+    "|": 4,
+    "é": 34,
+    "ó": 32,
+    "ŋ": 35,
+    "ɓ": 11,
+    "ɔ": 16,
+    "ɛ": 15,
+    "̀": 30,
+    "́": 33,
+    "̣": 36
+  },
+  "lsi": {
+    "'": 21,
+    "-": 27,
+    "0": 33,
+    "1": 35,
+    "2": 36,
+    "3": 41,
+    "4": 37,
+    "5": 38,
+    "6": 34,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 39,
+    "a": 7,
+    "b": 17,
+    "c": 29,
+    "d": 12,
+    "e": 14,
+    "f": 31,
+    "g": 8,
+    "h": 11,
+    "i": 6,
+    "j": 25,
+    "k": 18,
+    "l": 20,
+    "m": 15,
+    "n": 9,
+    "o": 5,
+    "p": 26,
+    "q": 32,
+    "r": 19,
+    "s": 16,
+    "t": 22,
+    "u": 13,
+    "v": 28,
+    "w": 23,
+    "x": 30,
+    "y": 10,
+    "z": 24,
+    "|": 4
+  },
+  "lsm": {
+    "'": 24,
+    "-": 27,
+    "0": 29,
+    "1": 30,
+    "2": 31,
+    "3": 34,
+    "4": 32,
+    "5": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "c": 23,
+    "d": 19,
+    "e": 7,
+    "f": 25,
+    "g": 22,
+    "h": 12,
+    "i": 6,
+    "j": 26,
+    "k": 20,
+    "l": 13,
+    "m": 14,
+    "n": 9,
+    "o": 8,
+    "p": 28,
+    "q": 35,
+    "r": 18,
+    "s": 16,
+    "t": 21,
+    "u": 10,
+    "w": 17,
+    "y": 15,
+    "|": 5
+  },
+  "ltz": {
+    "!": 76,
+    "\"": 61,
+    "$": 60,
+    "%": 58,
+    "&": 48,
+    "'": 30,
+    "+": 63,
+    ",": 51,
+    "-": 35,
+    ".": 45,
+    "/": 50,
+    "0": 32,
+    "1": 34,
+    "2": 36,
+    "3": 43,
+    "4": 41,
+    "5": 40,
+    "6": 44,
+    "7": 49,
+    "8": 42,
+    "9": 38,
+    ":": 53,
+    ";": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 72,
+    "]": 73,
+    "a": 9,
+    "b": 26,
+    "c": 18,
+    "d": 12,
+    "e": 5,
+    "f": 20,
+    "g": 17,
+    "h": 16,
+    "i": 10,
+    "j": 31,
+    "k": 22,
+    "l": 14,
+    "m": 19,
+    "n": 6,
+    "o": 15,
+    "p": 25,
+    "q": 47,
+    "r": 7,
+    "s": 11,
+    "t": 8,
+    "u": 13,
+    "v": 24,
+    "w": 21,
+    "x": 39,
+    "y": 33,
+    "z": 27,
+    "|": 4,
+    "¥": 65,
+    "°": 56,
+    "²": 62,
+    "á": 67,
+    "â": 69,
+    "ã": 68,
+    "ä": 29,
+    "æ": 64,
+    "ç": 66,
+    "è": 54,
+    "é": 23,
+    "ë": 28,
+    "í": 59,
+    "ï": 70,
+    "ó": 75,
+    "ö": 52,
+    "ú": 71,
+    "ü": 37,
+    "ł": 74,
+    "̇": 77,
+    "‑": 57,
+    "–": 55
+  },
+  "luc": {
+    "'": 10,
+    "-": 49,
+    "0": 56,
+    "1": 57,
+    "2": 59,
+    "3": 61,
+    "4": 60,
+    "6": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 16,
+    "c": 40,
+    "d": 8,
+    "e": 37,
+    "f": 45,
+    "g": 22,
+    "h": 50,
+    "i": 18,
+    "j": 27,
+    "k": 17,
+    "l": 15,
+    "m": 19,
+    "n": 20,
+    "o": 46,
+    "p": 30,
+    "r": 5,
+    "s": 24,
+    "t": 31,
+    "u": 28,
+    "v": 43,
+    "w": 42,
+    "y": 36,
+    "z": 32,
+    "|": 4,
+    "á": 6,
+    "â": 44,
+    "ã": 9,
+    "é": 25,
+    "ê": 54,
+    "í": 13,
+    "î": 48,
+    "ó": 26,
+    "ô": 51,
+    "õ": 39,
+    "ú": 34,
+    "û": 52,
+    "ĩ": 7,
+    "ŋ": 35,
+    "ũ": 33,
+    "́": 12,
+    "̂": 47,
+    "̃": 21,
+    "ẹ": 41,
+    "ẽ": 38,
+    "ệ": 55,
+    "ị": 11,
+    "ọ": 29,
+    "ộ": 53,
+    "ụ": 23
+  },
+  "lug": {
+    "!": 58,
+    "$": 52,
+    "%": 54,
+    "&": 66,
+    "'": 23,
+    "+": 55,
+    ",": 41,
+    "-": 44,
+    ".": 34,
+    "/": 48,
+    "0": 30,
+    "1": 31,
+    "2": 32,
+    "3": 40,
+    "4": 37,
+    "5": 36,
+    "6": 38,
+    "7": 43,
+    "8": 39,
+    "9": 35,
+    ":": 47,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 50,
+    "[": 70,
+    "]": 71,
+    "`": 73,
+    "a": 4,
+    "b": 12,
+    "c": 28,
+    "d": 22,
+    "e": 6,
+    "f": 25,
+    "g": 16,
+    "h": 29,
+    "i": 8,
+    "j": 24,
+    "k": 11,
+    "l": 14,
+    "m": 13,
+    "n": 10,
+    "o": 9,
+    "p": 26,
+    "q": 46,
+    "r": 20,
+    "s": 19,
+    "t": 18,
+    "u": 7,
+    "v": 27,
+    "w": 17,
+    "x": 45,
+    "y": 15,
+    "z": 21,
+    "|": 5,
+    "£": 59,
+    "¥": 64,
+    "°": 56,
+    "²": 69,
+    "á": 61,
+    "ã": 65,
+    "ç": 63,
+    "í": 62,
+    "õ": 57,
+    "ú": 68,
+    "ü": 49,
+    "ŋ": 67,
+    "̀": 78,
+    "́": 76,
+    "̓": 75,
+    "̔": 77,
+    "η": 74,
+    "ղ": 72,
+    "–": 51,
+    "—": 60,
+    "’": 33,
+    "”": 42
+  },
+  "luo": {
+    "$": 48,
+    "%": 51,
+    "&": 60,
+    "'": 26,
+    "+": 56,
+    ",": 43,
+    "-": 30,
+    ".": 40,
+    "/": 47,
+    "0": 28,
+    "1": 29,
+    "2": 32,
+    "3": 36,
+    "4": 35,
+    "5": 34,
+    "6": 38,
+    "7": 41,
+    "8": 39,
+    "9": 33,
+    ":": 45,
+    ";": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 58,
+    "]": 59,
+    "_": 66,
+    "a": 6,
+    "b": 23,
+    "c": 20,
+    "d": 14,
+    "e": 7,
+    "f": 27,
+    "g": 13,
+    "h": 16,
+    "i": 9,
+    "j": 25,
+    "k": 11,
+    "l": 18,
+    "m": 10,
+    "n": 8,
+    "o": 5,
+    "p": 24,
+    "q": 44,
+    "r": 12,
+    "s": 22,
+    "t": 15,
+    "u": 17,
+    "v": 31,
+    "w": 21,
+    "x": 42,
+    "y": 19,
+    "z": 37,
+    "|": 4,
+    "£": 67,
+    "¥": 54,
+    "°": 57,
+    "á": 49,
+    "í": 55,
+    "ó": 63,
+    "õ": 50,
+    "ú": 65,
+    "ü": 53,
+    "̇": 64,
+    "‘": 61,
+    "’": 62,
+    "”": 46
+  },
+  "lwo": {
+    "'": 25,
+    "0": 27,
+    "1": 29,
+    "5": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 21,
+    "d": 14,
+    "e": 6,
+    "g": 10,
+    "h": 8,
+    "i": 11,
+    "j": 20,
+    "k": 15,
+    "l": 23,
+    "m": 17,
+    "n": 7,
+    "o": 9,
+    "p": 24,
+    "q": 26,
+    "r": 16,
+    "s": 28,
+    "t": 22,
+    "u": 12,
+    "w": 18,
+    "y": 13,
+    "|": 4
+  },
+  "lww": {
+    "'": 34,
+    "-": 23,
+    "0": 26,
+    "1": 24,
+    "2": 25,
+    "3": 28,
+    "4": 27,
+    "5": 29,
+    "6": 30,
+    "7": 33,
+    "8": 32,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "e": 7,
+    "g": 19,
+    "i": 8,
+    "k": 16,
+    "l": 10,
+    "m": 14,
+    "n": 6,
+    "o": 12,
+    "p": 9,
+    "r": 15,
+    "s": 11,
+    "t": 18,
+    "u": 13,
+    "v": 20,
+    "w": 22,
+    "x": 35,
+    "y": 21,
+    "|": 5,
+    "̃": 17
+  },
+  "lzz": {
+    "'": 34,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "c": 31,
+    "d": 17,
+    "e": 6,
+    "f": 35,
+    "g": 22,
+    "h": 25,
+    "i": 7,
+    "j": 33,
+    "k": 12,
+    "l": 20,
+    "m": 11,
+    "n": 9,
+    "o": 8,
+    "p": 19,
+    "r": 14,
+    "s": 16,
+    "t": 15,
+    "u": 10,
+    "v": 24,
+    "x": 21,
+    "y": 26,
+    "z": 30,
+    "|": 4,
+    "ç": 23,
+    "ğ": 29,
+    "ş": 18,
+    "ʒ": 27,
+    "̆": 13,
+    "̇": 36
+  },
+  "maa-dialect_sanantonio": {
+    "'": 14,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 26,
+    "d": 27,
+    "e": 20,
+    "f": 37,
+    "g": 18,
+    "h": 25,
+    "i": 8,
+    "j": 9,
+    "k": 11,
+    "l": 21,
+    "m": 22,
+    "n": 6,
+    "o": 13,
+    "p": 39,
+    "q": 43,
+    "r": 19,
+    "s": 12,
+    "t": 10,
+    "u": 36,
+    "v": 41,
+    "x": 24,
+    "y": 28,
+    "z": 42,
+    "|": 4,
+    "à": 32,
+    "á": 17,
+    "è": 31,
+    "é": 29,
+    "ì": 23,
+    "í": 16,
+    "ñ": 33,
+    "ò": 30,
+    "ó": 34,
+    "ù": 44,
+    "ú": 38,
+    "̱": 7,
+    "‐": 15
+  },
+  "maa-dialect_sanjeronimo": {
+    "'": 14,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 27,
+    "d": 28,
+    "e": 16,
+    "f": 38,
+    "g": 18,
+    "h": 26,
+    "i": 8,
+    "j": 9,
+    "k": 10,
+    "l": 23,
+    "m": 25,
+    "n": 6,
+    "o": 12,
+    "p": 39,
+    "q": 42,
+    "r": 33,
+    "s": 13,
+    "t": 11,
+    "u": 36,
+    "v": 41,
+    "x": 17,
+    "y": 30,
+    "z": 40,
+    "|": 4,
+    "à": 22,
+    "á": 21,
+    "è": 29,
+    "é": 31,
+    "ì": 20,
+    "í": 19,
+    "ñ": 32,
+    "ò": 24,
+    "ó": 34,
+    "ú": 37,
+    "̱": 7,
+    "‐": 15
+  },
+  "mad": {
+    "'": 20,
+    "-": 25,
+    "0": 33,
+    "2": 36,
+    "3": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "c": 26,
+    "d": 15,
+    "e": 7,
+    "f": 28,
+    "g": 9,
+    "h": 21,
+    "i": 18,
+    "j": 24,
+    "k": 12,
+    "l": 13,
+    "m": 19,
+    "n": 6,
+    "o": 11,
+    "p": 17,
+    "r": 10,
+    "s": 8,
+    "t": 16,
+    "u": 22,
+    "w": 27,
+    "y": 23,
+    "z": 29,
+    "|": 5,
+    "â": 30,
+    "ä": 32,
+    "ë": 35,
+    "–": 31
+  },
+  "mag": {
+    "-": 53,
+    "2": 68,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "à": 67,
+    "è": 66,
+    "ʻ": 58,
+    "ʼ": 61,
+    "ँ": 40,
+    "ं": 32,
+    "अ": 30,
+    "आ": 23,
+    "इ": 28,
+    "ई": 39,
+    "उ": 20,
+    "ऊ": 42,
+    "ए": 45,
+    "ऐ": 62,
+    "ओ": 38,
+    "औ": 64,
+    "क": 7,
+    "ख": 31,
+    "ग": 27,
+    "घ": 54,
+    "च": 33,
+    "छ": 49,
+    "ज": 22,
+    "झ": 56,
+    "ञ": 41,
+    "ट": 47,
+    "ठ": 50,
+    "ड": 44,
+    "ढ": 51,
+    "ण": 55,
+    "त": 15,
+    "थ": 35,
+    "द": 24,
+    "ध": 46,
+    "न": 10,
+    "ऩ": 70,
+    "प": 18,
+    "फ": 48,
+    "ब": 19,
+    "भ": 37,
+    "म": 13,
+    "य": 21,
+    "र": 8,
+    "ऱ": 69,
+    "ल": 12,
+    "व": 29,
+    "श": 63,
+    "ष": 60,
+    "स": 11,
+    "ह": 9,
+    "़": 43,
+    "ऽ": 36,
+    "ा": 6,
+    "ि": 14,
+    "ी": 16,
+    "ु": 25,
+    "ू": 34,
+    "ॅ": 71,
+    "े": 5,
+    "ै": 57,
+    "ो": 17,
+    "ौ": 59,
+    "्": 26,
+    "‍": 52,
+    "–": 65
+  },
+  "mah": {
+    "'": 36,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 18,
+    "c": 39,
+    "d": 26,
+    "e": 5,
+    "f": 35,
+    "g": 31,
+    "h": 34,
+    "i": 7,
+    "j": 10,
+    "k": 9,
+    "l": 15,
+    "m": 13,
+    "n": 8,
+    "o": 11,
+    "p": 22,
+    "q": 40,
+    "r": 12,
+    "s": 27,
+    "t": 16,
+    "u": 24,
+    "v": 32,
+    "w": 17,
+    "x": 38,
+    "y": 30,
+    "z": 33,
+    "|": 4,
+    "ñ": 20,
+    "ā": 21,
+    "ō": 14,
+    "ū": 29,
+    "ḷ": 23,
+    "ṃ": 19,
+    "ṇ": 28,
+    "ọ": 25
+  },
+  "mai": {
+    "'": 66,
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 18,
+    "ं": 50,
+    "ः": 65,
+    "अ": 23,
+    "आ": 33,
+    "इ": 49,
+    "ई": 53,
+    "उ": 51,
+    "ऊ": 63,
+    "ऋ": 68,
+    "ए": 40,
+    "ऐ": 64,
+    "ओ": 34,
+    "औ": 52,
+    "क": 6,
+    "ख": 38,
+    "ग": 32,
+    "घ": 58,
+    "ङ": 67,
+    "च": 41,
+    "छ": 29,
+    "ज": 21,
+    "झ": 60,
+    "ञ": 61,
+    "ट": 43,
+    "ठ": 55,
+    "ड": 46,
+    "ढ": 59,
+    "ण": 48,
+    "त": 14,
+    "थ": 39,
+    "द": 27,
+    "ध": 42,
+    "न": 12,
+    "प": 17,
+    "फ": 56,
+    "ब": 25,
+    "भ": 20,
+    "म": 16,
+    "य": 24,
+    "र": 7,
+    "ल": 15,
+    "व": 28,
+    "श": 31,
+    "ष": 44,
+    "स": 11,
+    "ह": 9,
+    "़": 47,
+    "ऽ": 36,
+    "ा": 5,
+    "ि": 8,
+    "ी": 26,
+    "ु": 22,
+    "ू": 37,
+    "ृ": 57,
+    "े": 10,
+    "ै": 35,
+    "ो": 30,
+    "ौ": 54,
+    "्": 13,
+    "‍": 19,
+    "—": 62
+  },
+  "maj": {
+    "'": 13,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 25,
+    "c": 24,
+    "d": 27,
+    "e": 16,
+    "f": 34,
+    "g": 21,
+    "h": 14,
+    "i": 8,
+    "j": 12,
+    "k": 10,
+    "l": 33,
+    "m": 22,
+    "n": 6,
+    "o": 18,
+    "p": 32,
+    "r": 28,
+    "s": 15,
+    "t": 9,
+    "u": 11,
+    "v": 38,
+    "x": 17,
+    "y": 23,
+    "z": 36,
+    "|": 4,
+    "á": 20,
+    "é": 29,
+    "í": 19,
+    "ñ": 30,
+    "ó": 31,
+    "ú": 26,
+    "̱": 5,
+    "ṉ": 37,
+    "—": 35
+  },
+  "mak": {
+    "'": 17,
+    "-": 23,
+    "0": 30,
+    "1": 32,
+    "2": 33,
+    "4": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 25,
+    "d": 24,
+    "e": 14,
+    "f": 28,
+    "g": 8,
+    "h": 26,
+    "i": 7,
+    "j": 21,
+    "k": 9,
+    "l": 11,
+    "m": 10,
+    "n": 6,
+    "o": 18,
+    "p": 19,
+    "r": 15,
+    "s": 16,
+    "t": 12,
+    "u": 13,
+    "w": 27,
+    "y": 22,
+    "z": 29,
+    "|": 5
+  },
+  "mal": {
+    "!": 105,
+    "\"": 59,
+    "$": 106,
+    "%": 98,
+    "&": 122,
+    "'": 89,
+    "+": 113,
+    ",": 76,
+    "-": 62,
+    ".": 63,
+    "/": 86,
+    "0": 57,
+    "1": 60,
+    "2": 64,
+    "3": 74,
+    "4": 72,
+    "5": 71,
+    "6": 78,
+    "7": 80,
+    "8": 77,
+    "9": 70,
+    ":": 84,
+    ";": 94,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 109,
+    "[": 116,
+    "]": 117,
+    "a": 97,
+    "b": 100,
+    "c": 96,
+    "d": 121,
+    "e": 104,
+    "f": 115,
+    "g": 120,
+    "h": 111,
+    "i": 88,
+    "j": 129,
+    "k": 118,
+    "l": 112,
+    "m": 92,
+    "n": 103,
+    "o": 99,
+    "p": 95,
+    "q": 133,
+    "r": 101,
+    "s": 93,
+    "t": 102,
+    "u": 110,
+    "v": 128,
+    "w": 127,
+    "x": 114,
+    "y": 123,
+    "|": 5,
+    "°": 126,
+    "ം": 21,
+    "ഃ": 90,
+    "അ": 31,
+    "ആ": 45,
+    "ഇ": 51,
+    "ഈ": 67,
+    "ഉ": 49,
+    "ഊ": 85,
+    "ഋ": 124,
+    "എ": 40,
+    "ഏ": 61,
+    "ഐ": 82,
+    "ഒ": 44,
+    "ഓ": 68,
+    "ഔ": 83,
+    "ക": 7,
+    "ഖ": 69,
+    "ഗ": 35,
+    "ഘ": 73,
+    "ങ": 25,
+    "ച": 24,
+    "ഛ": 87,
+    "ജ": 41,
+    "ഞ": 46,
+    "ട": 15,
+    "ഠ": 79,
+    "ഡ": 47,
+    "ഢ": 119,
+    "ണ": 26,
+    "ത": 10,
+    "ഥ": 55,
+    "ദ": 34,
+    "ധ": 42,
+    "ന": 8,
+    "പ": 14,
+    "ഫ": 53,
+    "ബ": 43,
+    "ഭ": 48,
+    "മ": 18,
+    "യ": 12,
+    "ര": 13,
+    "റ": 19,
+    "ല": 20,
+    "ള": 23,
+    "ഴ": 52,
+    "വ": 17,
+    "ശ": 38,
+    "ഷ": 36,
+    "സ": 22,
+    "ഹ": 50,
+    "ാ": 11,
+    "ി": 6,
+    "ീ": 37,
+    "ു": 9,
+    "ൂ": 39,
+    "ൃ": 65,
+    "െ": 16,
+    "േ": 30,
+    "ൈ": 54,
+    "ൊ": 56,
+    "ോ": 27,
+    "ൌ": 91,
+    "്": 4,
+    "ൎ": 134,
+    "ൗ": 75,
+    "൪": 132,
+    "൯": 135,
+    "ൺ": 66,
+    "ൻ": 32,
+    "ർ": 29,
+    "ൽ": 28,
+    "ൾ": 33,
+    "‌": 58,
+    "‍": 81,
+    "–": 125,
+    "—": 131,
+    "‘": 107,
+    "’": 108,
+    "”": 130
+  },
+  "mam-dialect_central": {
+    "'": 6,
+    "-": 29,
+    "0": 39,
+    "1": 42,
+    "2": 44,
+    "3": 45,
+    "4": 41,
+    "5": 43,
+    "6": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 19,
+    "c": 24,
+    "d": 27,
+    "e": 11,
+    "f": 35,
+    "g": 31,
+    "h": 25,
+    "i": 7,
+    "j": 10,
+    "k": 12,
+    "l": 13,
+    "m": 20,
+    "n": 8,
+    "o": 16,
+    "p": 26,
+    "q": 18,
+    "r": 28,
+    "s": 23,
+    "t": 5,
+    "u": 15,
+    "v": 36,
+    "w": 22,
+    "x": 17,
+    "y": 14,
+    "z": 21,
+    "|": 4,
+    "á": 37,
+    "é": 33,
+    "í": 34,
+    "ñ": 46,
+    "ó": 38,
+    "ú": 30,
+    "ẍ": 32
+  },
+  "mam-dialect_northern": {
+    "-": 29,
+    "0": 41,
+    "1": 50,
+    "2": 43,
+    "3": 44,
+    "4": 42,
+    "5": 46,
+    "6": 47,
+    "7": 48,
+    "8": 49,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 11,
+    "d": 27,
+    "e": 8,
+    "f": 39,
+    "g": 32,
+    "h": 25,
+    "i": 9,
+    "j": 12,
+    "k": 18,
+    "l": 14,
+    "m": 19,
+    "n": 7,
+    "o": 17,
+    "p": 24,
+    "q": 28,
+    "r": 26,
+    "s": 22,
+    "t": 5,
+    "u": 13,
+    "v": 36,
+    "w": 23,
+    "x": 16,
+    "y": 15,
+    "z": 21,
+    "|": 4,
+    "á": 38,
+    "é": 34,
+    "í": 37,
+    "ñ": 40,
+    "ó": 35,
+    "ú": 30,
+    "ẍ": 33,
+    "—": 31,
+    "ꞌ": 10
+  },
+  "mam-dialect_southern": {
+    "'": 5,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 19,
+    "c": 24,
+    "d": 27,
+    "e": 9,
+    "g": 32,
+    "h": 25,
+    "i": 10,
+    "j": 11,
+    "k": 13,
+    "l": 14,
+    "m": 20,
+    "n": 8,
+    "o": 17,
+    "p": 26,
+    "q": 18,
+    "r": 28,
+    "s": 22,
+    "t": 6,
+    "u": 12,
+    "w": 23,
+    "x": 15,
+    "y": 16,
+    "z": 21,
+    "|": 4,
+    "ú": 33,
+    "́": 34,
+    "ẍ": 31,
+    "—": 30
+  },
+  "mam-dialect_western": {
+    "'": 10,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 19,
+    "c": 12,
+    "d": 25,
+    "e": 9,
+    "f": 39,
+    "g": 34,
+    "h": 27,
+    "i": 8,
+    "j": 11,
+    "k": 18,
+    "l": 13,
+    "m": 22,
+    "n": 7,
+    "o": 17,
+    "p": 24,
+    "q": 31,
+    "r": 26,
+    "s": 20,
+    "t": 5,
+    "u": 16,
+    "v": 40,
+    "w": 23,
+    "x": 14,
+    "y": 15,
+    "z": 21,
+    "|": 4,
+    "á": 36,
+    "é": 35,
+    "í": 33,
+    "ñ": 41,
+    "ó": 38,
+    "ú": 30,
+    "̈": 29,
+    "ẍ": 37,
+    "—": 32
+  },
+  "maq": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 9,
+    "d": 22,
+    "e": 12,
+    "f": 31,
+    "g": 20,
+    "h": 11,
+    "i": 8,
+    "j": 10,
+    "l": 27,
+    "m": 17,
+    "n": 6,
+    "o": 19,
+    "p": 29,
+    "q": 28,
+    "r": 18,
+    "s": 14,
+    "t": 13,
+    "u": 7,
+    "v": 23,
+    "x": 16,
+    "y": 21,
+    "z": 37,
+    "|": 4,
+    "á": 24,
+    "é": 33,
+    "ë": 15,
+    "í": 30,
+    "ñ": 25,
+    "ó": 36,
+    "ú": 26,
+    "́": 34,
+    "‍": 32
+  },
+  "mar": {
+    "!": 120,
+    "\"": 49,
+    "$": 114,
+    "%": 117,
+    "&": 138,
+    "'": 89,
+    "+": 123,
+    ",": 77,
+    "-": 66,
+    ".": 56,
+    "/": 91,
+    "0": 54,
+    "1": 55,
+    "2": 59,
+    "3": 71,
+    "4": 69,
+    "5": 72,
+    "6": 76,
+    "7": 82,
+    "8": 78,
+    "9": 70,
+    ":": 84,
+    ";": 115,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 135,
+    "[": 136,
+    "]": 137,
+    "a": 83,
+    "b": 110,
+    "c": 98,
+    "d": 101,
+    "e": 112,
+    "f": 107,
+    "g": 116,
+    "h": 108,
+    "i": 96,
+    "j": 126,
+    "k": 119,
+    "l": 113,
+    "m": 97,
+    "n": 100,
+    "o": 104,
+    "p": 95,
+    "q": 133,
+    "r": 111,
+    "s": 103,
+    "t": 102,
+    "u": 106,
+    "v": 105,
+    "w": 109,
+    "x": 122,
+    "y": 134,
+    "z": 129,
+    "|": 4,
+    "¥": 127,
+    "°": 124,
+    "²": 139,
+    "õ": 125,
+    "ʇ": 143,
+    "ʈ": 145,
+    "ँ": 68,
+    "ं": 21,
+    "ः": 79,
+    "अ": 30,
+    "आ": 25,
+    "इ": 47,
+    "ई": 58,
+    "उ": 45,
+    "ऊ": 61,
+    "ऋ": 118,
+    "ऍ": 132,
+    "ए": 43,
+    "ऐ": 93,
+    "ऑ": 64,
+    "ओ": 63,
+    "औ": 99,
+    "क": 11,
+    "ख": 39,
+    "ग": 29,
+    "घ": 46,
+    "ङ": 144,
+    "च": 22,
+    "छ": 74,
+    "ज": 27,
+    "झ": 50,
+    "ञ": 73,
+    "ट": 31,
+    "ठ": 40,
+    "ड": 36,
+    "ढ": 60,
+    "ण": 23,
+    "त": 9,
+    "थ": 41,
+    "द": 26,
+    "ध": 33,
+    "न": 15,
+    "प": 20,
+    "फ": 44,
+    "ब": 35,
+    "भ": 42,
+    "म": 19,
+    "य": 10,
+    "र": 7,
+    "ऱ": 57,
+    "ल": 14,
+    "ळ": 38,
+    "ऴ": 142,
+    "व": 17,
+    "श": 32,
+    "ष": 37,
+    "स": 13,
+    "ह": 18,
+    "़": 140,
+    "ा": 5,
+    "ि": 16,
+    "ी": 12,
+    "ु": 28,
+    "ू": 34,
+    "ृ": 52,
+    "ॅ": 51,
+    "े": 8,
+    "ै": 53,
+    "ॉ": 48,
+    "ॊ": 141,
+    "ो": 24,
+    "ौ": 67,
+    "्": 6,
+    "०": 62,
+    "१": 65,
+    "२": 80,
+    "३": 88,
+    "४": 92,
+    "५": 85,
+    "६": 86,
+    "७": 94,
+    "८": 87,
+    "९": 81,
+    "ॲ": 90,
+    "‍": 75,
+    "–": 128,
+    "—": 121,
+    "‘": 130,
+    "’": 131
+  },
+  "maw": {
+    "'": 28,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "d": 15,
+    "e": 19,
+    "f": 30,
+    "g": 22,
+    "h": 29,
+    "i": 6,
+    "k": 14,
+    "l": 13,
+    "m": 9,
+    "n": 7,
+    "o": 24,
+    "p": 26,
+    "r": 18,
+    "s": 12,
+    "t": 17,
+    "u": 8,
+    "v": 31,
+    "w": 27,
+    "y": 10,
+    "z": 25,
+    "|": 4,
+    "ŋ": 16,
+    "ɛ": 21,
+    "ԑ": 32,
+    "ↄ": 20
+  },
+  "maz": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 9,
+    "d": 20,
+    "e": 8,
+    "f": 43,
+    "g": 19,
+    "h": 27,
+    "i": 7,
+    "j": 6,
+    "k": 47,
+    "l": 35,
+    "m": 13,
+    "n": 12,
+    "o": 11,
+    "p": 22,
+    "q": 44,
+    "r": 15,
+    "s": 18,
+    "t": 17,
+    "u": 14,
+    "v": 33,
+    "x": 26,
+    "y": 21,
+    "z": 23,
+    "|": 4,
+    "á": 36,
+    "ã": 34,
+    "ä": 31,
+    "é": 45,
+    "ë": 32,
+    "í": 29,
+    "ñ": 28,
+    "ó": 38,
+    "õ": 40,
+    "ö": 24,
+    "ú": 42,
+    "ü": 16,
+    "ĩ": 46,
+    "ũ": 37,
+    "̱": 30,
+    "ṉ": 39,
+    "ẽ": 41
+  },
+  "mbb": {
+    "'": 33,
+    "-": 27,
+    "7": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 22,
+    "c": 34,
+    "d": 14,
+    "e": 5,
+    "f": 35,
+    "g": 16,
+    "h": 15,
+    "i": 8,
+    "j": 30,
+    "k": 11,
+    "l": 18,
+    "m": 13,
+    "n": 6,
+    "o": 32,
+    "p": 21,
+    "r": 26,
+    "s": 12,
+    "t": 10,
+    "u": 9,
+    "v": 24,
+    "w": 19,
+    "x": 36,
+    "y": 17,
+    "z": 20,
+    "|": 4,
+    "à": 23,
+    "è": 28,
+    "ì": 31,
+    "ù": 29,
+    "ǥ": 25
+  },
+  "mbc": {
+    "'": 9,
+    "-": 55,
+    "0": 33,
+    "1": 44,
+    "2": 43,
+    "3": 49,
+    "4": 47,
+    "5": 48,
+    "6": 52,
+    "7": 50,
+    "8": 54,
+    "9": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 24,
+    "d": 23,
+    "e": 10,
+    "f": 29,
+    "g": 30,
+    "h": 36,
+    "i": 16,
+    "j": 22,
+    "k": 15,
+    "l": 25,
+    "m": 14,
+    "n": 7,
+    "o": 12,
+    "p": 8,
+    "q": 34,
+    "r": 11,
+    "s": 19,
+    "t": 13,
+    "u": 18,
+    "v": 31,
+    "w": 20,
+    "x": 45,
+    "y": 17,
+    "z": 37,
+    "|": 4,
+    "á": 32,
+    "â": 53,
+    "ã": 28,
+    "é": 26,
+    "ê": 42,
+    "í": 38,
+    "î": 6,
+    "ó": 35,
+    "ô": 41,
+    "õ": 56,
+    "ú": 46,
+    "˻": 39,
+    "˼": 40,
+    "—": 21
+  },
+  "mbh": {
+    "'": 28,
+    "-": 36,
+    "0": 25,
+    "1": 26,
+    "2": 27,
+    "3": 29,
+    "4": 31,
+    "5": 30,
+    "6": 32,
+    "7": 35,
+    "8": 33,
+    "9": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "d": 21,
+    "e": 7,
+    "f": 38,
+    "g": 10,
+    "h": 20,
+    "i": 9,
+    "j": 23,
+    "k": 15,
+    "l": 11,
+    "m": 13,
+    "n": 8,
+    "o": 6,
+    "p": 18,
+    "r": 12,
+    "s": 19,
+    "t": 14,
+    "u": 16,
+    "v": 17,
+    "w": 22,
+    "y": 37,
+    "|": 4
+  },
+  "mbj": {
+    "'": 11,
+    "-": 35,
+    "0": 41,
+    "1": 39,
+    "2": 40,
+    "3": 46,
+    "4": 42,
+    "5": 44,
+    "6": 45,
+    "7": 43,
+    "8": 47,
+    "9": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "d": 8,
+    "e": 15,
+    "g": 22,
+    "h": 6,
+    "i": 24,
+    "j": 21,
+    "k": 16,
+    "l": 49,
+    "m": 17,
+    "n": 9,
+    "o": 7,
+    "p": 20,
+    "r": 25,
+    "s": 19,
+    "t": 10,
+    "u": 29,
+    "w": 26,
+    "y": 13,
+    "|": 4,
+    "ã": 23,
+    "ä": 14,
+    "é": 30,
+    "ë": 18,
+    "ó": 31,
+    "õ": 32,
+    "ĩ": 36,
+    "ũ": 34,
+    "ǟ": 38,
+    "̃": 37,
+    "ẽ": 28,
+    "ỹ": 27,
+    "—": 33
+  },
+  "mbt": {
+    "'": 25,
+    "-": 23,
+    "0": 26,
+    "1": 30,
+    "2": 28,
+    "3": 29,
+    "4": 32,
+    "5": 31,
+    "6": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "d": 16,
+    "e": 5,
+    "g": 12,
+    "h": 21,
+    "i": 8,
+    "k": 10,
+    "l": 17,
+    "m": 13,
+    "n": 7,
+    "p": 18,
+    "r": 20,
+    "s": 14,
+    "t": 11,
+    "u": 9,
+    "w": 19,
+    "y": 15,
+    "|": 4,
+    "—": 24
+  },
+  "mbu": {
+    "'": 33,
+    "-": 32,
+    "0": 56,
+    "1": 61,
+    "2": 63,
+    "4": 60,
+    "6": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 26,
+    "d": 21,
+    "e": 17,
+    "f": 38,
+    "g": 13,
+    "h": 30,
+    "i": 10,
+    "j": 36,
+    "k": 9,
+    "l": 12,
+    "m": 8,
+    "n": 6,
+    "o": 19,
+    "p": 24,
+    "r": 14,
+    "s": 16,
+    "t": 22,
+    "u": 11,
+    "v": 51,
+    "w": 18,
+    "y": 23,
+    "z": 31,
+    "|": 4,
+    "à": 25,
+    "á": 35,
+    "â": 50,
+    "è": 45,
+    "é": 48,
+    "ê": 53,
+    "ì": 29,
+    "í": 44,
+    "î": 47,
+    "ò": 62,
+    "ó": 55,
+    "ô": 58,
+    "ù": 43,
+    "ú": 37,
+    "û": 52,
+    "ā": 39,
+    "ě": 54,
+    "ŋ": 34,
+    "ū": 42,
+    "ǎ": 46,
+    "ǒ": 57,
+    "ǔ": 49,
+    "ǝ": 7,
+    "ɓ": 15,
+    "ɗ": 28,
+    "ə": 40,
+    "̀": 27,
+    "́": 41
+  },
+  "mbz": {
+    "-": 12,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 16,
+    "d": 32,
+    "e": 21,
+    "f": 33,
+    "g": 27,
+    "h": 17,
+    "i": 7,
+    "j": 24,
+    "k": 10,
+    "l": 29,
+    "m": 25,
+    "n": 6,
+    "o": 13,
+    "p": 30,
+    "r": 14,
+    "s": 18,
+    "t": 9,
+    "u": 8,
+    "v": 20,
+    "w": 23,
+    "x": 26,
+    "y": 15,
+    "|": 4,
+    "é": 35,
+    "í": 31,
+    "ñ": 22,
+    "ó": 36,
+    "ɨ": 19,
+    "—": 28,
+    "ꞌ": 11
+  },
+  "mca": {
+    "'": 12,
+    "-": 40,
+    "0": 44,
+    "1": 42,
+    "3": 45,
+    "4": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 32,
+    "c": 30,
+    "d": 29,
+    "e": 6,
+    "f": 26,
+    "g": 33,
+    "h": 16,
+    "i": 5,
+    "j": 11,
+    "k": 13,
+    "l": 23,
+    "m": 22,
+    "n": 10,
+    "o": 24,
+    "p": 20,
+    "q": 14,
+    "r": 27,
+    "s": 15,
+    "t": 9,
+    "u": 17,
+    "v": 39,
+    "w": 21,
+    "x": 25,
+    "y": 19,
+    "z": 37,
+    "|": 4,
+    "á": 38,
+    "é": 35,
+    "í": 34,
+    "ñ": 41,
+    "ó": 36,
+    "ú": 31,
+    "ƚ": 18,
+    "ˈ": 8,
+    "—": 28
+  },
+  "mcb": {
+    "'": 27,
+    "0": 25,
+    "1": 26,
+    "2": 28,
+    "3": 32,
+    "4": 29,
+    "5": 31,
+    "6": 33,
+    "7": 30,
+    "8": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 22,
+    "e": 11,
+    "g": 13,
+    "h": 19,
+    "i": 5,
+    "j": 23,
+    "k": 9,
+    "m": 15,
+    "n": 10,
+    "o": 12,
+    "p": 16,
+    "r": 7,
+    "s": 14,
+    "t": 8,
+    "u": 20,
+    "v": 17,
+    "y": 18,
+    "|": 6,
+    "ñ": 21,
+    "—": 24
+  },
+  "mcd": {
+    "'": 41,
+    "0": 34,
+    "1": 38,
+    "2": 43,
+    "4": 37,
+    "6": 39,
+    "7": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 11,
+    "d": 20,
+    "e": 21,
+    "f": 13,
+    "g": 28,
+    "h": 12,
+    "i": 7,
+    "j": 23,
+    "k": 36,
+    "l": 25,
+    "m": 15,
+    "n": 6,
+    "o": 8,
+    "p": 18,
+    "q": 19,
+    "r": 16,
+    "s": 10,
+    "t": 14,
+    "u": 9,
+    "v": 31,
+    "x": 35,
+    "y": 17,
+    "z": 33,
+    "|": 5,
+    "á": 30,
+    "é": 27,
+    "í": 29,
+    "ñ": 40,
+    "ó": 32,
+    "ú": 24,
+    "—": 22
+  },
+  "mco": {
+    "'": 18,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 23,
+    "c": 9,
+    "d": 14,
+    "e": 19,
+    "f": 42,
+    "g": 22,
+    "h": 17,
+    "i": 12,
+    "j": 8,
+    "k": 45,
+    "l": 31,
+    "m": 13,
+    "n": 10,
+    "o": 16,
+    "p": 21,
+    "q": 40,
+    "r": 28,
+    "s": 24,
+    "t": 11,
+    "u": 15,
+    "v": 44,
+    "x": 20,
+    "y": 6,
+    "z": 25,
+    "|": 4,
+    "á": 38,
+    "ä": 27,
+    "é": 39,
+    "ë": 34,
+    "í": 36,
+    "ï": 37,
+    "ñ": 26,
+    "ó": 35,
+    "ö": 30,
+    "ú": 33,
+    "ü": 29,
+    "ɨ": 5,
+    "́": 41,
+    "̈": 32
+  },
+  "mcp": {
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 9,
+    "c": 30,
+    "d": 16,
+    "e": 42,
+    "f": 41,
+    "g": 11,
+    "h": 28,
+    "i": 29,
+    "j": 27,
+    "k": 25,
+    "l": 17,
+    "m": 8,
+    "n": 7,
+    "o": 32,
+    "p": 39,
+    "r": 43,
+    "s": 13,
+    "t": 33,
+    "u": 26,
+    "v": 47,
+    "w": 10,
+    "y": 15,
+    "z": 34,
+    "|": 4,
+    "á": 12,
+    "â": 36,
+    "é": 35,
+    "ê": 48,
+    "í": 19,
+    "î": 37,
+    "ó": 40,
+    "ô": 46,
+    "ú": 20,
+    "û": 44,
+    "ŋ": 23,
+    "ǎ": 51,
+    "ǐ": 53,
+    "ǒ": 52,
+    "ǔ": 50,
+    "ɔ": 22,
+    "ə": 5,
+    "ɛ": 18,
+    "ɨ": 24,
+    "ʉ": 21,
+    "́": 6,
+    "̂": 31,
+    "̌": 49,
+    "̧": 38
+  },
+  "mcq": {
+    "'": 9,
+    "-": 35,
+    "0": 28,
+    "1": 32,
+    "2": 30,
+    "3": 36,
+    "5": 31,
+    "6": 33,
+    "7": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 25,
+    "d": 24,
+    "e": 7,
+    "g": 23,
+    "h": 11,
+    "i": 6,
+    "j": 12,
+    "k": 16,
+    "m": 15,
+    "n": 8,
+    "o": 13,
+    "p": 18,
+    "r": 14,
+    "s": 19,
+    "t": 21,
+    "u": 10,
+    "v": 17,
+    "w": 20,
+    "|": 5,
+    "á": 26,
+    "é": 29,
+    "í": 27,
+    "ú": 22
+  },
+  "mcu": {
+    "'": 55,
+    "-": 46,
+    "0": 60,
+    "1": 50,
+    "2": 51,
+    "3": 52,
+    "4": 54,
+    "5": 53,
+    "6": 57,
+    "7": 56,
+    "8": 59,
+    "9": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 47,
+    "a": 11,
+    "b": 9,
+    "c": 32,
+    "d": 13,
+    "e": 7,
+    "f": 38,
+    "g": 19,
+    "h": 22,
+    "i": 23,
+    "j": 35,
+    "k": 20,
+    "l": 14,
+    "m": 10,
+    "n": 5,
+    "o": 33,
+    "p": 48,
+    "r": 26,
+    "s": 15,
+    "t": 18,
+    "u": 27,
+    "v": 42,
+    "w": 31,
+    "y": 17,
+    "|": 4,
+    "à": 16,
+    "á": 25,
+    "â": 34,
+    "è": 12,
+    "é": 6,
+    "ê": 39,
+    "ì": 40,
+    "í": 30,
+    "î": 44,
+    "ò": 28,
+    "ó": 41,
+    "ô": 43,
+    "ù": 36,
+    "ú": 37,
+    "û": 49,
+    "ŋ": 21,
+    "ɔ": 8,
+    "̀": 29,
+    "́": 24,
+    "̂": 45
+  },
+  "mda": {
+    "-": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 58,
+    "a": 17,
+    "b": 16,
+    "c": 39,
+    "d": 18,
+    "e": 25,
+    "f": 42,
+    "g": 9,
+    "h": 36,
+    "i": 29,
+    "j": 38,
+    "k": 11,
+    "l": 22,
+    "m": 10,
+    "n": 6,
+    "o": 32,
+    "p": 30,
+    "r": 15,
+    "s": 13,
+    "t": 20,
+    "u": 23,
+    "v": 41,
+    "w": 21,
+    "y": 8,
+    "z": 33,
+    "|": 4,
+    "à": 26,
+    "â": 49,
+    "è": 35,
+    "é": 56,
+    "ê": 52,
+    "ì": 28,
+    "î": 55,
+    "ò": 45,
+    "ô": 54,
+    "ù": 37,
+    "û": 53,
+    "ā": 19,
+    "ă": 48,
+    "ē": 12,
+    "ĕ": 47,
+    "ī": 34,
+    "ĭ": 51,
+    "ō": 44,
+    "ŏ": 57,
+    "ū": 31,
+    "ŭ": 46,
+    "ɔ": 27,
+    "ə": 5,
+    "ɛ": 24,
+    "̀": 14,
+    "̂": 43,
+    "̄": 7,
+    "̆": 40
+  },
+  "mdf": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 29,
+    "в": 19,
+    "г": 23,
+    "д": 18,
+    "е": 9,
+    "ж": 35,
+    "з": 20,
+    "и": 13,
+    "й": 28,
+    "к": 12,
+    "л": 14,
+    "м": 15,
+    "н": 6,
+    "о": 11,
+    "п": 25,
+    "р": 17,
+    "с": 8,
+    "т": 7,
+    "у": 21,
+    "ф": 26,
+    "х": 27,
+    "ц": 24,
+    "ч": 36,
+    "ш": 22,
+    "щ": 38,
+    "ы": 33,
+    "ь": 10,
+    "э": 30,
+    "ю": 37,
+    "я": 16,
+    "ё": 31,
+    "—": 32
+  },
+  "mdv": {
+    "-": 12,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 23,
+    "d": 11,
+    "e": 7,
+    "f": 38,
+    "g": 34,
+    "h": 24,
+    "i": 8,
+    "j": 13,
+    "k": 10,
+    "l": 20,
+    "m": 27,
+    "n": 6,
+    "o": 16,
+    "p": 31,
+    "r": 26,
+    "s": 19,
+    "t": 15,
+    "u": 9,
+    "v": 21,
+    "x": 25,
+    "y": 18,
+    "z": 40,
+    "|": 4,
+    "á": 37,
+    "é": 33,
+    "í": 30,
+    "ñ": 22,
+    "ó": 39,
+    "ú": 28,
+    "ɨ": 17,
+    "́": 36,
+    "̱": 32,
+    "—": 29,
+    "ꞌ": 14
+  },
+  "mdy": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ሁ": 108,
+    "ሂ": 94,
+    "ሃ": 22,
+    "ሄ": 84,
+    "ሆ": 136,
+    "ሉ": 121,
+    "ሊ": 61,
+    "ላ": 35,
+    "ሌ": 78,
+    "ል": 100,
+    "ሎ": 56,
+    "ሙ": 53,
+    "ሚ": 68,
+    "ማ": 10,
+    "ሜ": 73,
+    "ም": 32,
+    "ሞ": 69,
+    "ሡ": 132,
+    "ሢ": 7,
+    "ሣ": 52,
+    "ሤ": 125,
+    "ሥ": 137,
+    "ሦ": 116,
+    "ሩ": 115,
+    "ሪ": 58,
+    "ራ": 67,
+    "ሬ": 113,
+    "ር": 24,
+    "ሮ": 43,
+    "ሱ": 62,
+    "ሲ": 26,
+    "ሳ": 48,
+    "ሴ": 29,
+    "ስ": 59,
+    "ሶ": 60,
+    "ሹ": 105,
+    "ሺ": 63,
+    "ሻ": 96,
+    "ሼ": 95,
+    "ሽ": 93,
+    "ሾ": 90,
+    "ቁ": 120,
+    "ቂ": 70,
+    "ቃ": 64,
+    "ቄ": 77,
+    "ቆ": 104,
+    "ቡ": 107,
+    "ቢ": 54,
+    "ባ": 28,
+    "ቤ": 72,
+    "ብ": 129,
+    "ቦ": 91,
+    "ቱ": 81,
+    "ቲ": 65,
+    "ታ": 15,
+    "ቴ": 19,
+    "ት": 155,
+    "ቶ": 31,
+    "ቹ": 139,
+    "ቺ": 111,
+    "ቻ": 119,
+    "ቼ": 140,
+    "ቾ": 114,
+    "ኑ": 57,
+    "ኒ": 47,
+    "ና": 8,
+    "ኔ": 17,
+    "ን": 5,
+    "ኖ": 89,
+    "ኡ": 76,
+    "ኢ": 74,
+    "ኣ": 13,
+    "ኤ": 37,
+    "ኦ": 39,
+    "ኩ": 97,
+    "ኪ": 44,
+    "ካ": 25,
+    "ኬ": 23,
+    "ክ": 151,
+    "ኮ": 18,
+    "ዉ": 131,
+    "ዊ": 141,
+    "ዋ": 36,
+    "ዌ": 135,
+    "ው": 101,
+    "ዎ": 38,
+    "ዑ": 80,
+    "ዒ": 9,
+    "ዓ": 6,
+    "ዔ": 16,
+    "ዖ": 42,
+    "ዙ": 127,
+    "ዚ": 50,
+    "ዛ": 20,
+    "ዜ": 87,
+    "ዝ": 153,
+    "ዞ": 71,
+    "ዡ": 152,
+    "ዢ": 126,
+    "ዣ": 128,
+    "ዤ": 134,
+    "ዥ": 150,
+    "ዦ": 99,
+    "ዩ": 148,
+    "ዪ": 130,
+    "ያ": 11,
+    "ዬ": 30,
+    "ይ": 12,
+    "ዮ": 102,
+    "ዱ": 92,
+    "ዲ": 46,
+    "ዳ": 14,
+    "ዴ": 49,
+    "ዶ": 66,
+    "ዹ": 117,
+    "ዺ": 82,
+    "ዻ": 55,
+    "ዼ": 79,
+    "ዾ": 106,
+    "ጁ": 154,
+    "ጂ": 138,
+    "ጃ": 147,
+    "ጄ": 146,
+    "ጆ": 142,
+    "ጉ": 41,
+    "ጊ": 45,
+    "ጋ": 34,
+    "ጌ": 21,
+    "ግ": 145,
+    "ጎ": 33,
+    "ጩ": 133,
+    "ጪ": 112,
+    "ጫ": 88,
+    "ጬ": 124,
+    "ጮ": 86,
+    "ጱ": 149,
+    "ጲ": 118,
+    "ጳ": 110,
+    "ጴ": 109,
+    "ጶ": 122,
+    "ፁ": 144,
+    "ፂ": 98,
+    "ፃ": 75,
+    "ፄ": 123,
+    "ፆ": 27,
+    "ፑ": 103,
+    "ፒ": 83,
+    "ፓ": 40,
+    "ፔ": 51,
+    "ፕ": 143,
+    "ፖ": 85
+  },
+  "med": {
+    "'": 30,
+    "-": 33,
+    "0": 26,
+    "1": 27,
+    "2": 29,
+    "3": 35,
+    "4": 31,
+    "5": 32,
+    "6": 34,
+    "7": 28,
+    "8": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 22,
+    "d": 19,
+    "e": 12,
+    "g": 16,
+    "i": 6,
+    "j": 24,
+    "k": 9,
+    "l": 18,
+    "m": 8,
+    "n": 5,
+    "o": 15,
+    "p": 14,
+    "r": 17,
+    "s": 23,
+    "t": 13,
+    "u": 11,
+    "w": 20,
+    "y": 25,
+    "|": 4,
+    "ʉ": 38,
+    "ᵾ": 10,
+    "ⱡ": 21
+  },
+  "mee": {
+    "'": 21,
+    "0": 22,
+    "2": 25,
+    "3": 26,
+    "4": 23,
+    "5": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "e": 6,
+    "g": 9,
+    "i": 11,
+    "k": 8,
+    "l": 10,
+    "m": 14,
+    "n": 7,
+    "o": 13,
+    "p": 17,
+    "r": 15,
+    "s": 18,
+    "t": 16,
+    "u": 12,
+    "v": 20,
+    "|": 5,
+    "—": 27
+  },
+  "mej": {
+    "'": 28,
+    "-": 27,
+    "0": 30,
+    "1": 31,
+    "2": 34,
+    "4": 32,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 25,
+    "d": 19,
+    "e": 7,
+    "f": 16,
+    "g": 11,
+    "h": 21,
+    "i": 8,
+    "j": 18,
+    "k": 15,
+    "l": 22,
+    "m": 14,
+    "n": 10,
+    "o": 6,
+    "p": 26,
+    "r": 9,
+    "s": 13,
+    "t": 17,
+    "u": 12,
+    "w": 24,
+    "y": 23,
+    "z": 29,
+    "|": 4
+  },
+  "men": {
+    "'": 32,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 33,
+    "d": 24,
+    "e": 8,
+    "f": 27,
+    "g": 10,
+    "h": 21,
+    "i": 6,
+    "j": 25,
+    "k": 18,
+    "l": 9,
+    "m": 14,
+    "n": 7,
+    "o": 19,
+    "p": 23,
+    "r": 28,
+    "s": 22,
+    "t": 17,
+    "u": 12,
+    "v": 26,
+    "w": 15,
+    "y": 16,
+    "z": 31,
+    "|": 4,
+    "ŋ": 30,
+    "ɔ": 13,
+    "ɛ": 11
+  },
+  "meq": {
+    "-": 35,
+    "0": 36,
+    "2": 39,
+    "3": 38,
+    "4": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 16,
+    "e": 6,
+    "f": 27,
+    "g": 19,
+    "h": 14,
+    "i": 15,
+    "k": 12,
+    "l": 13,
+    "m": 9,
+    "n": 10,
+    "o": 23,
+    "p": 29,
+    "r": 18,
+    "s": 20,
+    "t": 11,
+    "u": 22,
+    "v": 28,
+    "w": 17,
+    "y": 7,
+    "z": 24,
+    "|": 4,
+    "â": 30,
+    "ŋ": 26,
+    "œ": 33,
+    "ɓ": 31,
+    "ɗ": 25,
+    "ə": 8,
+    "ɨ": 32,
+    "ʼ": 34
+  },
+  "met": {
+    "'": 34,
+    "-": 38,
+    "0": 25,
+    "1": 26,
+    "2": 30,
+    "3": 33,
+    "4": 31,
+    "5": 35,
+    "6": 32,
+    "7": 28,
+    "8": 36,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 10,
+    "d": 13,
+    "e": 20,
+    "g": 9,
+    "h": 19,
+    "i": 7,
+    "k": 22,
+    "l": 12,
+    "m": 11,
+    "n": 6,
+    "o": 17,
+    "p": 21,
+    "r": 18,
+    "s": 16,
+    "t": 14,
+    "u": 8,
+    "w": 24,
+    "x": 15,
+    "y": 23,
+    "|": 5,
+    "ʼ": 27,
+    "—": 29
+  },
+  "mev": {
+    "'": 52,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 12,
+    "b": 32,
+    "d": 31,
+    "e": 15,
+    "f": 42,
+    "g": 22,
+    "h": 49,
+    "i": 21,
+    "k": 8,
+    "l": 7,
+    "m": 14,
+    "n": 30,
+    "o": 17,
+    "p": 27,
+    "s": 34,
+    "t": 37,
+    "u": 44,
+    "v": 43,
+    "w": 19,
+    "y": 24,
+    "z": 26,
+    "|": 4,
+    "à": 10,
+    "á": 13,
+    "ã": 39,
+    "è": 25,
+    "é": 18,
+    "ì": 23,
+    "í": 20,
+    "ò": 33,
+    "ó": 29,
+    "ù": 41,
+    "ú": 40,
+    "ĩ": 36,
+    "ŋ": 50,
+    "ũ": 45,
+    "ƃ": 48,
+    "ɓ": 35,
+    "ɔ": 9,
+    "ɛ": 5,
+    "̀": 6,
+    "́": 11,
+    "̃": 16,
+    "ḿ": 46,
+    "ṹ": 47,
+    "‐": 38,
+    "—": 51
+  },
+  "mfe": {
+    "'": 28,
+    "-": 26,
+    "0": 29,
+    "1": 31,
+    "2": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "d": 19,
+    "e": 7,
+    "f": 21,
+    "g": 24,
+    "h": 30,
+    "i": 8,
+    "k": 14,
+    "l": 12,
+    "m": 17,
+    "n": 5,
+    "o": 9,
+    "p": 15,
+    "r": 10,
+    "s": 13,
+    "t": 11,
+    "u": 18,
+    "v": 22,
+    "w": 23,
+    "x": 27,
+    "y": 25,
+    "z": 16,
+    "|": 4,
+    "–": 33
+  },
+  "mfh": {
+    "'": 33,
+    "-": 38,
+    "1": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 40,
+    "d": 22,
+    "e": 25,
+    "f": 30,
+    "g": 10,
+    "h": 21,
+    "i": 20,
+    "j": 39,
+    "k": 8,
+    "l": 6,
+    "m": 12,
+    "n": 19,
+    "o": 23,
+    "p": 29,
+    "r": 28,
+    "s": 15,
+    "t": 16,
+    "u": 11,
+    "v": 27,
+    "w": 13,
+    "y": 18,
+    "z": 14,
+    "|": 5,
+    "à": 7,
+    "è": 37,
+    "ì": 36,
+    "ò": 34,
+    "ù": 35,
+    "ŋ": 17,
+    "ɓ": 32,
+    "ɗ": 26,
+    "ə": 9,
+    "̀": 31
+  },
+  "mfi": {
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 31,
+    "d": 13,
+    "e": 6,
+    "f": 28,
+    "g": 21,
+    "h": 22,
+    "i": 10,
+    "j": 30,
+    "k": 12,
+    "l": 19,
+    "m": 11,
+    "n": 7,
+    "p": 29,
+    "r": 9,
+    "s": 16,
+    "t": 14,
+    "u": 15,
+    "v": 27,
+    "w": 18,
+    "y": 17,
+    "z": 24,
+    "|": 5,
+    "á": 8,
+    "í": 34,
+    "ú": 35,
+    "ŋ": 23,
+    "ƴ": 33,
+    "ɓ": 32,
+    "ɗ": 25,
+    "ꞌ": 36
+  },
+  "mfk": {
+    "'": 29,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 34,
+    "d": 11,
+    "e": 8,
+    "f": 27,
+    "g": 21,
+    "h": 17,
+    "i": 24,
+    "k": 10,
+    "l": 16,
+    "m": 6,
+    "n": 9,
+    "o": 28,
+    "p": 30,
+    "r": 15,
+    "s": 18,
+    "t": 14,
+    "u": 22,
+    "v": 26,
+    "w": 13,
+    "y": 7,
+    "z": 19,
+    "|": 5,
+    "ä": 33,
+    "ŋ": 25,
+    "ɓ": 31,
+    "ɗ": 23,
+    "ə": 12
+  },
+  "mfq": {
+    "'": 22,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 28,
+    "b": 10,
+    "c": 27,
+    "d": 19,
+    "e": 13,
+    "f": 25,
+    "g": 26,
+    "h": 30,
+    "i": 7,
+    "j": 24,
+    "k": 8,
+    "l": 12,
+    "m": 14,
+    "n": 6,
+    "o": 17,
+    "p": 20,
+    "s": 15,
+    "t": 11,
+    "u": 9,
+    "w": 23,
+    "y": 16,
+    "|": 4,
+    "ŋ": 18,
+    "ɑ": 5,
+    "ɔ": 21
+  },
+  "mfy": {
+    "'": 43,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 23,
+    "d": 31,
+    "e": 6,
+    "f": 38,
+    "g": 34,
+    "h": 15,
+    "i": 10,
+    "j": 14,
+    "k": 9,
+    "l": 25,
+    "m": 11,
+    "n": 12,
+    "o": 13,
+    "p": 21,
+    "q": 42,
+    "r": 18,
+    "s": 17,
+    "t": 7,
+    "u": 8,
+    "v": 39,
+    "x": 44,
+    "y": 26,
+    "z": 41,
+    "|": 4,
+    "á": 19,
+    "ä": 22,
+    "é": 20,
+    "ë": 29,
+    "í": 24,
+    "ï": 28,
+    "ñ": 37,
+    "ó": 32,
+    "ö": 33,
+    "ú": 30,
+    "ü": 27,
+    "‐": 35,
+    "—": 36
+  },
+  "mfz": {
+    "'": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 15,
+    "d": 22,
+    "e": 10,
+    "g": 20,
+    "h": 30,
+    "i": 8,
+    "j": 24,
+    "k": 9,
+    "l": 19,
+    "m": 17,
+    "n": 7,
+    "o": 14,
+    "p": 26,
+    "r": 25,
+    "t": 13,
+    "u": 23,
+    "w": 21,
+    "y": 12,
+    "|": 4,
+    "ç": 32,
+    "ñ": 27,
+    "ŋ": 16,
+    "š": 33,
+    "ɔ": 11,
+    "ɛ": 6,
+    "ḍ": 29,
+    "ṭ": 28
+  },
+  "mgd": {
+    "'": 19,
+    "-": 41,
+    "2": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 29,
+    "d": 13,
+    "e": 8,
+    "f": 32,
+    "g": 20,
+    "h": 34,
+    "i": 7,
+    "j": 30,
+    "k": 14,
+    "l": 21,
+    "m": 16,
+    "n": 9,
+    "o": 6,
+    "p": 26,
+    "r": 10,
+    "s": 18,
+    "t": 11,
+    "u": 15,
+    "v": 24,
+    "w": 28,
+    "y": 12,
+    "z": 23,
+    "|": 4,
+    "à": 27,
+    "ä": 22,
+    "è": 35,
+    "ì": 33,
+    "í": 38,
+    "î": 36,
+    "ò": 37,
+    "ù": 39,
+    "ŋ": 25,
+    "̀": 40,
+    "ṛ": 31
+  },
+  "mge": {
+    "-": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 28,
+    "d": 10,
+    "e": 21,
+    "g": 22,
+    "i": 14,
+    "j": 15,
+    "k": 7,
+    "l": 13,
+    "m": 20,
+    "n": 8,
+    "o": 18,
+    "p": 34,
+    "r": 19,
+    "s": 26,
+    "t": 12,
+    "u": 25,
+    "w": 32,
+    "y": 29,
+    "|": 4,
+    "à": 37,
+    "á": 36,
+    "è": 33,
+    "é": 16,
+    "ò": 38,
+    "ó": 39,
+    "ù": 46,
+    "ú": 43,
+    "ɓ": 31,
+    "ɔ": 17,
+    "ɗ": 44,
+    "ə": 11,
+    "ɛ": 41,
+    "ɨ": 5,
+    "̀": 35,
+    "́": 24,
+    "̰": 30,
+    "ḛ": 27,
+    "ḭ": 40,
+    "ṵ": 42,
+    "–": 45,
+    "ꞌ": 23
+  },
+  "mgh": {
+    "'": 29,
+    "-": 34,
+    "0": 33,
+    "1": 30,
+    "2": 31,
+    "6": 32,
+    "8": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 19,
+    "d": 25,
+    "e": 11,
+    "f": 26,
+    "g": 23,
+    "h": 6,
+    "i": 7,
+    "j": 28,
+    "k": 12,
+    "l": 14,
+    "m": 16,
+    "n": 8,
+    "o": 10,
+    "p": 20,
+    "r": 18,
+    "s": 21,
+    "t": 13,
+    "u": 9,
+    "v": 22,
+    "w": 15,
+    "y": 17,
+    "z": 27,
+    "|": 5
+  },
+  "mgo": {
+    "'": 44,
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 11,
+    "b": 9,
+    "c": 27,
+    "d": 23,
+    "e": 12,
+    "f": 30,
+    "g": 17,
+    "h": 19,
+    "i": 16,
+    "j": 34,
+    "k": 22,
+    "l": 42,
+    "m": 8,
+    "n": 7,
+    "o": 33,
+    "p": 41,
+    "r": 20,
+    "s": 26,
+    "t": 24,
+    "u": 29,
+    "v": 46,
+    "w": 14,
+    "y": 15,
+    "z": 28,
+    "|": 4,
+    "à": 25,
+    "â": 49,
+    "è": 37,
+    "ê": 52,
+    "ì": 32,
+    "î": 36,
+    "ò": 40,
+    "ô": 48,
+    "ù": 38,
+    "û": 35,
+    "ě": 47,
+    "ŋ": 18,
+    "ǎ": 43,
+    "ǐ": 51,
+    "ǒ": 50,
+    "ǔ": 53,
+    "ɔ": 21,
+    "ə": 6,
+    "ɨ": 5,
+    "ʼ": 13,
+    "̀": 10,
+    "̂": 31,
+    "̌": 39
+  },
+  "mhi": {
+    "'": 17,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 27,
+    "d": 12,
+    "e": 11,
+    "f": 29,
+    "g": 21,
+    "h": 28,
+    "i": 5,
+    "j": 22,
+    "k": 13,
+    "l": 14,
+    "m": 19,
+    "n": 9,
+    "o": 8,
+    "p": 25,
+    "r": 7,
+    "s": 15,
+    "t": 18,
+    "u": 10,
+    "v": 24,
+    "w": 26,
+    "y": 20,
+    "z": 23,
+    "|": 4,
+    "—": 31
+  },
+  "mhr": {
+    "'": 41,
+    "-": 28,
+    "1": 44,
+    "4": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 37,
+    "в": 24,
+    "г": 21,
+    "д": 15,
+    "е": 8,
+    "ж": 25,
+    "з": 29,
+    "и": 18,
+    "й": 20,
+    "к": 12,
+    "л": 11,
+    "м": 10,
+    "н": 7,
+    "о": 14,
+    "п": 23,
+    "р": 17,
+    "с": 19,
+    "т": 9,
+    "у": 16,
+    "ф": 38,
+    "х": 35,
+    "ц": 40,
+    "ч": 22,
+    "ш": 13,
+    "щ": 42,
+    "ъ": 39,
+    "ы": 6,
+    "ь": 34,
+    "э": 33,
+    "ю": 32,
+    "я": 31,
+    "ҥ": 30,
+    "ӧ": 27,
+    "ӱ": 26,
+    "–": 36
+  },
+  "mhu": {
+    "'": 33,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 27,
+    "d": 22,
+    "e": 7,
+    "g": 10,
+    "h": 6,
+    "i": 13,
+    "j": 30,
+    "k": 16,
+    "l": 15,
+    "m": 14,
+    "n": 9,
+    "o": 23,
+    "p": 29,
+    "q": 28,
+    "r": 24,
+    "s": 20,
+    "t": 19,
+    "u": 18,
+    "v": 36,
+    "w": 12,
+    "y": 8,
+    "z": 32,
+    "|": 4,
+    "ã": 26,
+    "ü": 11,
+    "ũ": 17,
+    "ǃ": 34,
+    "̃": 35,
+    "ẽ": 25
+  },
+  "mhx": {
+    "'": 17,
+    "-": 25,
+    "0": 33,
+    "1": 35,
+    "2": 36,
+    "4": 34,
+    "6": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 30,
+    "d": 16,
+    "e": 9,
+    "f": 32,
+    "g": 7,
+    "h": 11,
+    "i": 15,
+    "j": 26,
+    "k": 18,
+    "l": 21,
+    "m": 13,
+    "n": 8,
+    "o": 6,
+    "p": 27,
+    "q": 23,
+    "r": 14,
+    "s": 20,
+    "t": 24,
+    "u": 10,
+    "v": 29,
+    "w": 22,
+    "x": 31,
+    "y": 12,
+    "z": 28,
+    "|": 4
+  },
+  "mhy": {
+    "-": 25,
+    "0": 29,
+    "1": 31,
+    "2": 33,
+    "3": 34,
+    "4": 30,
+    "5": 35,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 28,
+    "d": 21,
+    "e": 9,
+    "f": 26,
+    "g": 17,
+    "h": 13,
+    "i": 8,
+    "j": 23,
+    "k": 12,
+    "l": 10,
+    "m": 16,
+    "n": 6,
+    "o": 24,
+    "p": 19,
+    "r": 11,
+    "s": 15,
+    "t": 14,
+    "u": 7,
+    "w": 20,
+    "y": 18,
+    "z": 27,
+    "|": 5
+  },
+  "mib": {
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 40,
+    "c": 9,
+    "d": 16,
+    "e": 8,
+    "f": 43,
+    "g": 37,
+    "h": 11,
+    "i": 10,
+    "j": 12,
+    "k": 44,
+    "l": 36,
+    "m": 30,
+    "n": 5,
+    "o": 21,
+    "p": 38,
+    "q": 31,
+    "r": 22,
+    "s": 17,
+    "t": 13,
+    "u": 6,
+    "v": 29,
+    "x": 39,
+    "y": 15,
+    "z": 42,
+    "|": 4,
+    "á": 14,
+    "é": 26,
+    "í": 18,
+    "ñ": 28,
+    "ó": 27,
+    "ú": 20,
+    "ā": 24,
+    "ē": 23,
+    "ī": 33,
+    "ō": 32,
+    "ū": 25,
+    "ɨ": 19,
+    "́": 34,
+    "̄": 35,
+    "‐": 41
+  },
+  "mie": {
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 15,
+    "b": 37,
+    "c": 6,
+    "d": 7,
+    "e": 16,
+    "f": 40,
+    "g": 38,
+    "h": 10,
+    "i": 9,
+    "j": 13,
+    "k": 41,
+    "l": 33,
+    "m": 27,
+    "n": 5,
+    "o": 24,
+    "p": 34,
+    "q": 29,
+    "r": 31,
+    "s": 20,
+    "t": 19,
+    "u": 12,
+    "v": 25,
+    "x": 36,
+    "y": 23,
+    "z": 39,
+    "|": 4,
+    "á": 11,
+    "é": 30,
+    "í": 17,
+    "ñ": 26,
+    "ó": 28,
+    "ú": 18,
+    "ā": 8,
+    "ē": 22,
+    "ī": 14,
+    "ō": 32,
+    "ū": 21
+  },
+  "mif": {
+    "'": 33,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 27,
+    "d": 10,
+    "e": 7,
+    "f": 25,
+    "g": 15,
+    "h": 18,
+    "i": 28,
+    "j": 30,
+    "k": 14,
+    "l": 13,
+    "m": 8,
+    "n": 16,
+    "p": 29,
+    "r": 12,
+    "s": 17,
+    "t": 19,
+    "u": 24,
+    "v": 23,
+    "w": 11,
+    "y": 6,
+    "z": 22,
+    "|": 5,
+    "ŋ": 20,
+    "ɓ": 32,
+    "ɗ": 26,
+    "ə": 9
+  },
+  "mih": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 30,
+    "d": 16,
+    "e": 20,
+    "f": 35,
+    "g": 34,
+    "h": 36,
+    "i": 8,
+    "j": 21,
+    "k": 10,
+    "l": 29,
+    "m": 24,
+    "n": 7,
+    "o": 13,
+    "p": 33,
+    "q": 39,
+    "r": 17,
+    "s": 11,
+    "t": 6,
+    "u": 9,
+    "v": 18,
+    "x": 28,
+    "y": 12,
+    "z": 37,
+    "|": 4,
+    "á": 26,
+    "é": 31,
+    "í": 25,
+    "ñ": 15,
+    "ó": 27,
+    "ú": 23,
+    "ɨ": 19,
+    "́": 38,
+    "̱": 22,
+    "ꞌ": 14
+  },
+  "mil": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 48,
+    "c": 8,
+    "d": 6,
+    "e": 16,
+    "f": 50,
+    "g": 25,
+    "h": 10,
+    "i": 12,
+    "j": 41,
+    "k": 52,
+    "l": 44,
+    "m": 40,
+    "n": 5,
+    "o": 26,
+    "p": 47,
+    "q": 33,
+    "r": 36,
+    "s": 22,
+    "t": 13,
+    "u": 7,
+    "v": 32,
+    "x": 19,
+    "y": 21,
+    "z": 51,
+    "|": 4,
+    "à": 23,
+    "á": 9,
+    "è": 30,
+    "é": 20,
+    "ì": 29,
+    "í": 15,
+    "ñ": 24,
+    "ò": 45,
+    "ó": 28,
+    "ù": 31,
+    "ú": 17,
+    "ü": 37,
+    "ě": 38,
+    "ǎ": 27,
+    "ǐ": 35,
+    "ǒ": 39,
+    "ǔ": 42,
+    "ɨ": 18,
+    "̀": 43,
+    "́": 34,
+    "̌": 49,
+    "‐": 11,
+    "—": 46
+  },
+  "mim": {
+    "'": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 33,
+    "c": 24,
+    "d": 13,
+    "e": 25,
+    "f": 36,
+    "g": 34,
+    "h": 26,
+    "i": 8,
+    "j": 29,
+    "k": 11,
+    "l": 31,
+    "m": 28,
+    "n": 6,
+    "o": 9,
+    "p": 32,
+    "q": 38,
+    "r": 23,
+    "s": 14,
+    "t": 12,
+    "u": 21,
+    "v": 20,
+    "x": 19,
+    "y": 22,
+    "z": 37,
+    "|": 4,
+    "á": 15,
+    "é": 30,
+    "í": 16,
+    "ñ": 17,
+    "ó": 18,
+    "ú": 27,
+    "̱": 7,
+    "‐": 35,
+    "ꞌ": 10
+  },
+  "min": {
+    "'": 29,
+    "-": 20,
+    "0": 31,
+    "1": 34,
+    "2": 35,
+    "3": 37,
+    "4": 32,
+    "5": 36,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 26,
+    "d": 13,
+    "e": 23,
+    "f": 27,
+    "g": 18,
+    "h": 19,
+    "i": 8,
+    "j": 24,
+    "k": 7,
+    "l": 17,
+    "m": 12,
+    "n": 6,
+    "o": 10,
+    "p": 21,
+    "r": 15,
+    "s": 14,
+    "t": 11,
+    "u": 9,
+    "w": 25,
+    "y": 22,
+    "z": 28,
+    "|": 5,
+    "—": 30
+  },
+  "mio": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 7,
+    "d": 13,
+    "e": 18,
+    "f": 28,
+    "g": 27,
+    "h": 10,
+    "i": 6,
+    "j": 23,
+    "k": 35,
+    "l": 24,
+    "m": 21,
+    "n": 8,
+    "o": 12,
+    "p": 25,
+    "q": 20,
+    "r": 14,
+    "s": 17,
+    "t": 11,
+    "u": 9,
+    "v": 15,
+    "x": 22,
+    "y": 19,
+    "z": 29,
+    "|": 4,
+    "á": 37,
+    "é": 36,
+    "í": 33,
+    "ñ": 16,
+    "ó": 34,
+    "̄": 31,
+    "ρ": 30
+  },
+  "mip": {
+    "-": 24,
+    "5": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 19,
+    "d": 10,
+    "e": 15,
+    "f": 36,
+    "g": 25,
+    "h": 20,
+    "i": 7,
+    "j": 21,
+    "k": 11,
+    "l": 27,
+    "m": 23,
+    "n": 6,
+    "o": 14,
+    "p": 29,
+    "q": 39,
+    "r": 22,
+    "s": 8,
+    "t": 12,
+    "u": 9,
+    "v": 26,
+    "x": 16,
+    "y": 17,
+    "z": 38,
+    "|": 4,
+    "á": 30,
+    "é": 35,
+    "í": 28,
+    "ñ": 18,
+    "ó": 37,
+    "ú": 33,
+    "ü": 41,
+    "̱": 40,
+    "̶": 31,
+    "—": 34,
+    "ꞌ": 13
+  },
+  "miq": {
+    "'": 28,
+    "-": 25,
+    "0": 29,
+    "1": 30,
+    "2": 32,
+    "3": 35,
+    "4": 33,
+    "5": 34,
+    "6": 36,
+    "7": 31,
+    "8": 39,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 9,
+    "d": 18,
+    "e": 23,
+    "f": 41,
+    "g": 21,
+    "h": 19,
+    "i": 6,
+    "j": 24,
+    "k": 8,
+    "l": 13,
+    "m": 14,
+    "n": 7,
+    "o": 27,
+    "p": 17,
+    "r": 10,
+    "s": 11,
+    "t": 15,
+    "u": 12,
+    "v": 42,
+    "w": 16,
+    "y": 20,
+    "|": 5,
+    "â": 22,
+    "ê": 40,
+    "î": 26,
+    "ô": 43,
+    "û": 37
+  },
+  "mit": {
+    "0": 53,
+    "3": 51,
+    "4": 52,
+    "9": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 38,
+    "c": 8,
+    "d": 11,
+    "e": 18,
+    "f": 41,
+    "g": 34,
+    "h": 12,
+    "i": 6,
+    "j": 37,
+    "k": 50,
+    "l": 33,
+    "m": 23,
+    "n": 5,
+    "o": 21,
+    "p": 35,
+    "q": 31,
+    "r": 29,
+    "s": 15,
+    "t": 17,
+    "u": 10,
+    "v": 22,
+    "x": 24,
+    "y": 26,
+    "z": 42,
+    "|": 4,
+    "à": 9,
+    "á": 16,
+    "â": 39,
+    "è": 25,
+    "é": 30,
+    "ê": 44,
+    "ì": 14,
+    "í": 20,
+    "î": 40,
+    "ñ": 27,
+    "ò": 36,
+    "ó": 32,
+    "ô": 47,
+    "ù": 19,
+    "ú": 28,
+    "û": 46,
+    "ü": 48,
+    "ˊ": 45,
+    "ˋ": 43,
+    "‐": 13
+  },
+  "miy": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 36,
+    "c": 23,
+    "d": 13,
+    "e": 11,
+    "f": 39,
+    "g": 34,
+    "h": 24,
+    "i": 7,
+    "j": 21,
+    "k": 12,
+    "l": 29,
+    "m": 26,
+    "n": 8,
+    "o": 15,
+    "p": 32,
+    "q": 41,
+    "r": 22,
+    "s": 17,
+    "t": 14,
+    "u": 10,
+    "v": 19,
+    "x": 16,
+    "y": 20,
+    "z": 40,
+    "|": 4,
+    "­": 43,
+    "á": 27,
+    "ä": 33,
+    "é": 38,
+    "ë": 42,
+    "í": 25,
+    "ï": 35,
+    "ñ": 18,
+    "ó": 30,
+    "ö": 37,
+    "ú": 28,
+    "ü": 31,
+    "̱": 5,
+    "ꞌ": 9
+  },
+  "miz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 39,
+    "c": 37,
+    "d": 19,
+    "e": 13,
+    "f": 44,
+    "g": 42,
+    "h": 40,
+    "i": 8,
+    "j": 33,
+    "k": 9,
+    "l": 36,
+    "m": 30,
+    "n": 5,
+    "o": 14,
+    "p": 38,
+    "q": 48,
+    "r": 31,
+    "s": 16,
+    "t": 7,
+    "u": 10,
+    "v": 17,
+    "x": 21,
+    "y": 50,
+    "z": 47,
+    "|": 4,
+    "á": 15,
+    "â": 45,
+    "é": 18,
+    "ê": 51,
+    "í": 20,
+    "î": 46,
+    "ñ": 11,
+    "ó": 27,
+    "ô": 49,
+    "ú": 24,
+    "û": 43,
+    "ā": 23,
+    "ē": 29,
+    "ī": 26,
+    "ō": 25,
+    "ū": 28,
+    "ɨ": 22,
+    "ʼ": 41,
+    "́": 32,
+    "̂": 52,
+    "̄": 34,
+    "—": 35,
+    "ꞌ": 12
+  },
+  "mjl": {
+    "'": 59,
+    "-": 53,
+    "0": 64,
+    "2": 63,
+    "7": 65,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 45,
+    "ं": 17,
+    "ः": 56,
+    "अ": 44,
+    "आ": 27,
+    "इ": 47,
+    "ई": 31,
+    "उ": 49,
+    "ऊ": 43,
+    "ए": 35,
+    "ऐ": 46,
+    "ओ": 57,
+    "औ": 62,
+    "क": 11,
+    "ख": 33,
+    "ग": 28,
+    "घ": 55,
+    "च": 32,
+    "छ": 51,
+    "ज": 20,
+    "झ": 37,
+    "ञ": 58,
+    "ट": 50,
+    "ठ": 41,
+    "ड": 36,
+    "ढ": 54,
+    "ण": 26,
+    "त": 15,
+    "थ": 39,
+    "द": 24,
+    "ध": 42,
+    "न": 16,
+    "प": 22,
+    "फ": 48,
+    "ब": 25,
+    "भ": 34,
+    "म": 18,
+    "य": 21,
+    "र": 7,
+    "ल": 19,
+    "व": 30,
+    "ष": 60,
+    "स": 9,
+    "ह": 10,
+    "़": 38,
+    "ा": 5,
+    "ि": 13,
+    "ी": 12,
+    "ु": 23,
+    "ू": 29,
+    "ृ": 61,
+    "े": 6,
+    "ै": 40,
+    "ो": 14,
+    "ौ": 52,
+    "्": 8
+  },
+  "mjv": {
+    "'": 48,
+    "-": 56,
+    "0": 55,
+    "3": 54,
+    "4": 53,
+    "6": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ം": 21,
+    "അ": 25,
+    "ആ": 40,
+    "ഇ": 35,
+    "ഈ": 49,
+    "ഉ": 41,
+    "ഊ": 47,
+    "എ": 34,
+    "ഏ": 39,
+    "ഒ": 32,
+    "ഓ": 44,
+    "ക": 6,
+    "ങ": 33,
+    "ച": 20,
+    "ജ": 51,
+    "ഞ": 42,
+    "ട": 13,
+    "ണ": 23,
+    "ത": 7,
+    "ന": 12,
+    "പ": 17,
+    "ബ": 43,
+    "മ": 19,
+    "യ": 16,
+    "ര": 18,
+    "റ": 30,
+    "ല": 15,
+    "ള": 24,
+    "വ": 14,
+    "ശ": 27,
+    "ാ": 11,
+    "ി": 8,
+    "ീ": 36,
+    "ു": 10,
+    "ൂ": 37,
+    "െ": 9,
+    "േ": 22,
+    "ൊ": 29,
+    "ോ": 26,
+    "്": 4,
+    "ൺ": 38,
+    "ൻ": 28,
+    "ർ": 46,
+    "ൽ": 31,
+    "ൾ": 45,
+    "‍": 50
+  },
+  "mkd": {
+    "!": 94,
+    "\"": 74,
+    "$": 83,
+    "%": 77,
+    "&": 88,
+    "'": 76,
+    "+": 95,
+    ",": 60,
+    "-": 36,
+    ".": 41,
+    "/": 63,
+    "0": 31,
+    "1": 33,
+    "2": 35,
+    "3": 43,
+    "4": 40,
+    "5": 38,
+    "6": 44,
+    "7": 46,
+    "8": 42,
+    "9": 37,
+    ":": 62,
+    ";": 80,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 92,
+    "]": 93,
+    "`": 89,
+    "a": 45,
+    "b": 71,
+    "c": 54,
+    "d": 55,
+    "e": 47,
+    "f": 69,
+    "g": 67,
+    "h": 61,
+    "i": 50,
+    "j": 81,
+    "k": 68,
+    "l": 56,
+    "m": 53,
+    "n": 51,
+    "o": 49,
+    "p": 59,
+    "q": 85,
+    "r": 52,
+    "s": 48,
+    "t": 57,
+    "u": 64,
+    "v": 70,
+    "w": 75,
+    "x": 78,
+    "y": 72,
+    "z": 84,
+    "|": 4,
+    "¥": 82,
+    "°": 90,
+    "²": 87,
+    "è": 86,
+    "õ": 79,
+    "ú": 96,
+    "а": 5,
+    "б": 23,
+    "в": 13,
+    "г": 21,
+    "д": 14,
+    "е": 8,
+    "ж": 27,
+    "з": 20,
+    "и": 7,
+    "к": 15,
+    "л": 16,
+    "м": 18,
+    "н": 10,
+    "о": 6,
+    "п": 17,
+    "р": 11,
+    "с": 12,
+    "т": 9,
+    "у": 19,
+    "ф": 29,
+    "х": 32,
+    "ц": 26,
+    "ч": 24,
+    "ш": 25,
+    "ѐ": 58,
+    "ѓ": 34,
+    "ѕ": 65,
+    "ј": 22,
+    "љ": 66,
+    "њ": 28,
+    "ќ": 30,
+    "ѝ": 73,
+    "џ": 39,
+    "—": 91
+  },
+  "mkl": {
+    "1": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 34,
+    "b": 14,
+    "c": 29,
+    "d": 18,
+    "e": 12,
+    "f": 28,
+    "g": 27,
+    "h": 32,
+    "i": 6,
+    "j": 24,
+    "k": 11,
+    "l": 21,
+    "m": 13,
+    "n": 8,
+    "o": 17,
+    "p": 31,
+    "r": 30,
+    "s": 16,
+    "t": 26,
+    "u": 10,
+    "v": 43,
+    "w": 19,
+    "y": 20,
+    "z": 36,
+    "|": 4,
+    "à": 42,
+    "á": 45,
+    "è": 50,
+    "é": 47,
+    "ì": 38,
+    "í": 23,
+    "ò": 40,
+    "ó": 48,
+    "ù": 35,
+    "ú": 44,
+    "ĩ": 39,
+    "ń": 46,
+    "ŋ": 15,
+    "ũ": 41,
+    "ǹ": 37,
+    "ɑ": 5,
+    "ɔ": 9,
+    "ɛ": 7,
+    "̀": 25,
+    "́": 33,
+    "̃": 22
+  },
+  "mkn": {
+    "'": 28,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 25,
+    "d": 12,
+    "e": 15,
+    "f": 32,
+    "g": 8,
+    "h": 23,
+    "i": 7,
+    "j": 22,
+    "k": 19,
+    "l": 17,
+    "m": 16,
+    "n": 6,
+    "o": 11,
+    "p": 18,
+    "r": 20,
+    "s": 10,
+    "t": 13,
+    "u": 9,
+    "w": 26,
+    "y": 21,
+    "|": 4,
+    "á": 27,
+    "é": 30,
+    "í": 33,
+    "ó": 31,
+    "ú": 29
+  },
+  "mlg": {
+    "'": 21,
+    "-": 25,
+    "0": 34,
+    "1": 35,
+    "2": 36,
+    "3": 43,
+    "4": 38,
+    "5": 40,
+    "6": 41,
+    "7": 44,
+    "8": 42,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 45,
+    "d": 19,
+    "e": 13,
+    "f": 17,
+    "g": 27,
+    "h": 14,
+    "i": 7,
+    "j": 26,
+    "k": 16,
+    "l": 20,
+    "m": 12,
+    "n": 6,
+    "o": 9,
+    "p": 23,
+    "r": 10,
+    "s": 15,
+    "t": 11,
+    "v": 22,
+    "y": 8,
+    "z": 18,
+    "|": 5,
+    "à": 28,
+    "è": 37,
+    "é": 46,
+    "ì": 33,
+    "î": 32,
+    "ò": 31,
+    "ô": 29,
+    "ỳ": 30,
+    "–": 47
+  },
+  "mlt": {
+    "!": 69,
+    "\"": 54,
+    "$": 56,
+    "%": 57,
+    "&": 78,
+    "'": 31,
+    "+": 65,
+    ",": 50,
+    "-": 15,
+    ".": 48,
+    "/": 53,
+    "0": 35,
+    "1": 37,
+    "2": 41,
+    "3": 45,
+    "4": 44,
+    "5": 43,
+    "6": 47,
+    "7": 49,
+    "8": 46,
+    "9": 42,
+    ":": 52,
+    ";": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 71,
+    "]": 72,
+    "a": 5,
+    "b": 21,
+    "c": 36,
+    "d": 19,
+    "e": 9,
+    "f": 22,
+    "g": 24,
+    "h": 26,
+    "i": 6,
+    "j": 16,
+    "k": 18,
+    "l": 8,
+    "m": 14,
+    "n": 10,
+    "o": 17,
+    "p": 23,
+    "q": 30,
+    "r": 11,
+    "s": 13,
+    "t": 7,
+    "u": 12,
+    "v": 33,
+    "w": 25,
+    "x": 32,
+    "y": 40,
+    "z": 27,
+    "|": 4,
+    "£": 70,
+    "¥": 61,
+    "°": 73,
+    "²": 77,
+    "à": 38,
+    "á": 59,
+    "ã": 68,
+    "ç": 62,
+    "è": 60,
+    "é": 75,
+    "ì": 74,
+    "í": 63,
+    "ò": 67,
+    "ó": 82,
+    "õ": 66,
+    "ù": 80,
+    "ú": 76,
+    "ü": 55,
+    "ċ": 34,
+    "ġ": 28,
+    "ħ": 20,
+    "ł": 79,
+    "ż": 29,
+    "–": 81,
+    "—": 64,
+    "’": 39,
+    "”": 51
+  },
+  "mmg": {
+    "0": 27,
+    "1": 26,
+    "2": 28,
+    "3": 33,
+    "4": 31,
+    "5": 30,
+    "6": 29,
+    "8": 32,
+    "9": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 16,
+    "e": 5,
+    "f": 22,
+    "g": 10,
+    "h": 19,
+    "i": 15,
+    "j": 25,
+    "k": 23,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 17,
+    "r": 7,
+    "s": 20,
+    "t": 9,
+    "u": 13,
+    "v": 24,
+    "w": 21,
+    "y": 12,
+    "|": 4,
+    "ō": 18
+  },
+  "mnb": {
+    "'": 17,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "d": 15,
+    "e": 9,
+    "f": 23,
+    "g": 22,
+    "h": 12,
+    "i": 8,
+    "k": 13,
+    "l": 18,
+    "m": 10,
+    "n": 7,
+    "o": 6,
+    "p": 20,
+    "r": 24,
+    "s": 16,
+    "t": 14,
+    "u": 11,
+    "w": 21,
+    "y": 26,
+    "z": 27,
+    "|": 5
+  },
+  "mnf": {
+    "'": 53,
+    "-": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 6,
+    "c": 49,
+    "d": 29,
+    "e": 10,
+    "f": 32,
+    "g": 23,
+    "h": 27,
+    "i": 24,
+    "j": 41,
+    "k": 18,
+    "l": 16,
+    "m": 11,
+    "n": 7,
+    "o": 33,
+    "p": 38,
+    "s": 21,
+    "t": 9,
+    "u": 19,
+    "v": 43,
+    "w": 20,
+    "y": 26,
+    "z": 34,
+    "|": 4,
+    "à": 8,
+    "â": 44,
+    "è": 14,
+    "ê": 31,
+    "ì": 35,
+    "î": 51,
+    "ò": 25,
+    "ô": 52,
+    "ù": 36,
+    "û": 42,
+    "ě": 46,
+    "ŋ": 28,
+    "ǎ": 39,
+    "ǐ": 50,
+    "ǒ": 48,
+    "ǔ": 47,
+    "ȩ": 40,
+    "ɔ": 13,
+    "ə": 30,
+    "ɨ": 12,
+    "ˈ": 22,
+    "̀": 17,
+    "̂": 37,
+    "̌": 45,
+    "̧": 15
+  },
+  "mnk": {
+    "'": 33,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 32,
+    "d": 20,
+    "e": 8,
+    "f": 22,
+    "h": 26,
+    "i": 7,
+    "j": 24,
+    "k": 12,
+    "l": 9,
+    "m": 15,
+    "n": 11,
+    "o": 6,
+    "p": 28,
+    "r": 19,
+    "s": 18,
+    "t": 13,
+    "u": 10,
+    "w": 21,
+    "y": 17,
+    "|": 4,
+    "à": 31,
+    "ì": 25,
+    "ñ": 23,
+    "ŋ": 14,
+    "ǹ": 29,
+    "̀": 30
+  },
+  "mnw": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 6,
+    "က": 10,
+    "ခ": 45,
+    "ဂ": 11,
+    "ဃ": 61,
+    "စ": 35,
+    "ဆ": 48,
+    "ဇ": 37,
+    "ည": 31,
+    "ဍ": 32,
+    "ဏ": 52,
+    "တ": 7,
+    "ထ": 50,
+    "ဒ": 21,
+    "ဓ": 53,
+    "န": 22,
+    "ပ": 14,
+    "ဖ": 51,
+    "ဗ": 40,
+    "ဘ": 54,
+    "မ": 9,
+    "ယ": 42,
+    "ရ": 24,
+    "လ": 25,
+    "ဝ": 29,
+    "သ": 23,
+    "ဟ": 27,
+    "အ": 28,
+    "ဣ": 56,
+    "ဥ": 57,
+    "ဨ": 55,
+    "ဩ": 58,
+    "ါ": 46,
+    "ာ": 12,
+    "ိ": 13,
+    "ီ": 34,
+    "ု": 5,
+    "ူ": 43,
+    "ေ": 17,
+    "ဲ": 16,
+    "ဳ": 44,
+    "ဴ": 47,
+    "ဵ": 30,
+    "ံ": 8,
+    "း": 26,
+    "္": 20,
+    "်": 4,
+    "ျ": 38,
+    "ြ": 41,
+    "ွ": 18,
+    "ှ": 15,
+    "ဿ": 60,
+    "၆": 62,
+    "ၚ": 19,
+    "ၜ": 49,
+    "ၝ": 59,
+    "ၞ": 33,
+    "ၟ": 36,
+    "ၠ": 39
+  },
+  "mnx": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 19,
+    "d": 12,
+    "e": 6,
+    "f": 26,
+    "g": 9,
+    "h": 18,
+    "i": 11,
+    "j": 24,
+    "k": 23,
+    "l": 17,
+    "m": 13,
+    "n": 7,
+    "o": 8,
+    "p": 25,
+    "r": 16,
+    "s": 14,
+    "t": 21,
+    "u": 10,
+    "w": 22,
+    "y": 20,
+    "|": 4
+  },
+  "moa": {
+    "'": 5,
+    "-": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 11,
+    "c": 33,
+    "d": 25,
+    "e": 9,
+    "f": 30,
+    "g": 24,
+    "h": 18,
+    "i": 12,
+    "j": 31,
+    "k": 14,
+    "l": 10,
+    "m": 17,
+    "n": 13,
+    "o": 16,
+    "p": 22,
+    "q": 34,
+    "r": 26,
+    "s": 27,
+    "t": 28,
+    "u": 20,
+    "v": 32,
+    "w": 21,
+    "y": 19,
+    "z": 23,
+    "|": 4,
+    "ŋ": 29,
+    "ɔ": 15,
+    "ɛ": 7
+  },
+  "mog": {
+    "'": 17,
+    "-": 28,
+    "0": 36,
+    "1": 39,
+    "4": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 33,
+    "d": 15,
+    "e": 21,
+    "f": 38,
+    "g": 13,
+    "h": 24,
+    "i": 8,
+    "j": 31,
+    "k": 12,
+    "l": 22,
+    "m": 10,
+    "n": 6,
+    "o": 7,
+    "p": 18,
+    "r": 20,
+    "s": 16,
+    "t": 9,
+    "u": 11,
+    "w": 27,
+    "y": 19,
+    "z": 35,
+    "|": 4,
+    "á": 23,
+    "é": 34,
+    "í": 25,
+    "ó": 32,
+    "ú": 30,
+    "ḷ": 26,
+    "ḻ": 29
+  },
+  "mon": {
+    "!": 87,
+    "\"": 71,
+    "%": 76,
+    "&": 95,
+    "'": 75,
+    "+": 85,
+    ",": 49,
+    "-": 40,
+    ".": 50,
+    "/": 54,
+    "0": 35,
+    "1": 37,
+    "2": 41,
+    "3": 45,
+    "4": 44,
+    "5": 43,
+    "6": 46,
+    "7": 48,
+    "8": 47,
+    "9": 42,
+    ":": 55,
+    ";": 84,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 72,
+    "[": 93,
+    "]": 94,
+    "a": 52,
+    "b": 77,
+    "c": 58,
+    "d": 68,
+    "e": 66,
+    "f": 74,
+    "g": 65,
+    "h": 70,
+    "i": 51,
+    "j": 86,
+    "k": 82,
+    "l": 64,
+    "m": 61,
+    "n": 63,
+    "o": 59,
+    "p": 53,
+    "q": 99,
+    "r": 56,
+    "s": 57,
+    "t": 60,
+    "u": 62,
+    "v": 67,
+    "w": 80,
+    "x": 83,
+    "y": 78,
+    "z": 91,
+    "|": 4,
+    "²": 92,
+    "õ": 81,
+    "а": 5,
+    "б": 19,
+    "в": 24,
+    "г": 8,
+    "д": 12,
+    "е": 29,
+    "ж": 23,
+    "з": 26,
+    "и": 13,
+    "й": 15,
+    "к": 32,
+    "л": 9,
+    "м": 22,
+    "н": 6,
+    "о": 10,
+    "п": 34,
+    "р": 11,
+    "с": 18,
+    "т": 17,
+    "у": 16,
+    "ф": 38,
+    "х": 14,
+    "ц": 30,
+    "ч": 27,
+    "ш": 31,
+    "щ": 97,
+    "ъ": 69,
+    "ы": 25,
+    "ь": 28,
+    "э": 7,
+    "ю": 36,
+    "я": 33,
+    "ё": 39,
+    "ѳ": 96,
+    "ү": 20,
+    "ө": 21,
+    "​": 79,
+    "–": 90,
+    "—": 98,
+    "‘": 88,
+    "’": 89,
+    "”": 73
+  },
+  "mop": {
+    "'": 6,
+    "-": 31,
+    "0": 35,
+    "1": 38,
+    "2": 37,
+    "3": 41,
+    "4": 44,
+    "5": 42,
+    "6": 43,
+    "7": 46,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 10,
+    "d": 25,
+    "e": 9,
+    "f": 32,
+    "g": 30,
+    "h": 20,
+    "i": 7,
+    "j": 15,
+    "k": 39,
+    "l": 13,
+    "m": 18,
+    "n": 14,
+    "o": 12,
+    "p": 23,
+    "q": 22,
+    "r": 28,
+    "s": 17,
+    "t": 11,
+    "u": 8,
+    "v": 33,
+    "w": 24,
+    "x": 21,
+    "y": 19,
+    "z": 27,
+    "|": 4,
+    "é": 34,
+    "ñ": 40,
+    "ü": 26,
+    "у": 36,
+    "—": 29
+  },
+  "mor": {
+    "'": 36,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 27,
+    "d": 22,
+    "e": 9,
+    "f": 29,
+    "g": 14,
+    "i": 6,
+    "j": 32,
+    "k": 30,
+    "l": 11,
+    "m": 18,
+    "n": 7,
+    "o": 16,
+    "p": 31,
+    "r": 12,
+    "s": 25,
+    "t": 28,
+    "u": 17,
+    "v": 35,
+    "w": 20,
+    "y": 26,
+    "|": 4,
+    "ë": 23,
+    "ñ": 21,
+    "đ": 13,
+    "ŋ": 10,
+    "ǝ": 8,
+    "ɽ": 19,
+    "ḏ": 33,
+    "ṯ": 15
+  },
+  "mos": {
+    "'": 35,
+    "-": 23,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 8,
+    "d": 17,
+    "e": 7,
+    "f": 30,
+    "g": 11,
+    "h": 36,
+    "i": 20,
+    "k": 19,
+    "l": 16,
+    "m": 9,
+    "n": 6,
+    "o": 12,
+    "p": 27,
+    "r": 21,
+    "s": 10,
+    "t": 15,
+    "u": 28,
+    "v": 34,
+    "w": 24,
+    "y": 13,
+    "z": 26,
+    "|": 4,
+    "ã": 14,
+    "õ": 29,
+    "ĩ": 33,
+    "ũ": 32,
+    "ɛ": 31,
+    "ɩ": 22,
+    "ʋ": 25,
+    "ẽ": 18,
+    "–": 37
+  },
+  "mox": {
+    "'": 27,
+    "-": 35,
+    "0": 26,
+    "1": 28,
+    "2": 30,
+    "3": 31,
+    "4": 29,
+    "5": 32,
+    "6": 34,
+    "7": 36,
+    "8": 37,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "d": 14,
+    "e": 7,
+    "f": 24,
+    "g": 15,
+    "h": 25,
+    "i": 6,
+    "k": 23,
+    "l": 16,
+    "m": 17,
+    "n": 8,
+    "o": 10,
+    "q": 21,
+    "s": 22,
+    "t": 11,
+    "u": 9,
+    "v": 13,
+    "w": 19,
+    "y": 20,
+    "|": 5,
+    "ʼ": 12,
+    "—": 38
+  },
+  "moz": {
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 33,
+    "d": 12,
+    "e": 6,
+    "g": 18,
+    "h": 30,
+    "i": 7,
+    "j": 34,
+    "k": 10,
+    "l": 22,
+    "m": 17,
+    "n": 9,
+    "o": 8,
+    "p": 26,
+    "r": 19,
+    "s": 20,
+    "t": 13,
+    "u": 11,
+    "w": 23,
+    "y": 14,
+    "z": 24,
+    "|": 4,
+    "á": 36,
+    "â": 40,
+    "í": 29,
+    "î": 37,
+    "ó": 32,
+    "ô": 38,
+    "ú": 35,
+    "û": 41,
+    "ŋ": 15,
+    "ƴ": 39,
+    "ɓ": 28,
+    "ɗ": 21,
+    "ʼ": 16,
+    "̰": 27,
+    "—": 42
+  },
+  "mpg": {
+    "'": 16,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 27,
+    "d": 10,
+    "e": 12,
+    "f": 31,
+    "g": 11,
+    "h": 19,
+    "i": 6,
+    "j": 26,
+    "k": 15,
+    "l": 13,
+    "m": 7,
+    "n": 8,
+    "o": 17,
+    "p": 28,
+    "q": 42,
+    "r": 21,
+    "s": 18,
+    "t": 14,
+    "u": 9,
+    "v": 25,
+    "w": 24,
+    "x": 44,
+    "y": 23,
+    "z": 22,
+    "|": 4,
+    "à": 39,
+    "á": 30,
+    "â": 29,
+    "ä": 34,
+    "ê": 35,
+    "ì": 40,
+    "î": 32,
+    "ï": 36,
+    "ô": 38,
+    "ö": 43,
+    "û": 37,
+    "ü": 41
+  },
+  "mpm": {
+    "'": 11,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 30,
+    "c": 19,
+    "d": 15,
+    "e": 21,
+    "f": 34,
+    "g": 27,
+    "h": 20,
+    "i": 7,
+    "j": 10,
+    "k": 9,
+    "l": 26,
+    "m": 25,
+    "n": 6,
+    "o": 14,
+    "p": 29,
+    "q": 38,
+    "r": 18,
+    "s": 16,
+    "t": 13,
+    "u": 8,
+    "v": 22,
+    "x": 24,
+    "y": 17,
+    "z": 36,
+    "|": 4,
+    "á": 35,
+    "é": 33,
+    "í": 32,
+    "ñ": 23,
+    "ó": 31,
+    "ú": 28,
+    "ā": 39,
+    "ī": 42,
+    "ū": 37,
+    "ɨ": 12,
+    "́": 41,
+    "̄": 40
+  },
+  "mpp": {
+    "'": 37,
+    "-": 24,
+    "0": 28,
+    "1": 27,
+    "2": 30,
+    "3": 35,
+    "4": 31,
+    "5": 32,
+    "6": 34,
+    "7": 29,
+    "8": 36,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 10,
+    "d": 19,
+    "e": 5,
+    "f": 23,
+    "g": 8,
+    "h": 22,
+    "i": 9,
+    "k": 18,
+    "l": 12,
+    "m": 13,
+    "n": 7,
+    "o": 11,
+    "p": 25,
+    "r": 26,
+    "s": 20,
+    "t": 21,
+    "u": 14,
+    "w": 17,
+    "y": 15,
+    "z": 33,
+    "|": 4,
+    "­": 39
+  },
+  "mpx": {
+    "'": 25,
+    "-": 24,
+    "0": 27,
+    "1": 28,
+    "2": 29,
+    "3": 32,
+    "4": 31,
+    "5": 33,
+    "6": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 15,
+    "d": 23,
+    "e": 8,
+    "g": 12,
+    "h": 18,
+    "i": 6,
+    "k": 17,
+    "l": 9,
+    "m": 19,
+    "n": 7,
+    "o": 10,
+    "p": 21,
+    "s": 20,
+    "t": 14,
+    "u": 11,
+    "v": 22,
+    "w": 13,
+    "y": 16,
+    "|": 5,
+    "—": 26
+  },
+  "mqb": {
+    "-": 43,
+    "2": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 28,
+    "d": 25,
+    "e": 10,
+    "f": 31,
+    "g": 24,
+    "h": 19,
+    "i": 15,
+    "j": 29,
+    "k": 9,
+    "l": 17,
+    "m": 11,
+    "n": 6,
+    "o": 16,
+    "p": 26,
+    "r": 20,
+    "s": 14,
+    "t": 12,
+    "u": 18,
+    "v": 30,
+    "w": 13,
+    "y": 8,
+    "z": 27,
+    "|": 4,
+    "à": 23,
+    "á": 40,
+    "â": 34,
+    "é": 41,
+    "ì": 38,
+    "î": 42,
+    "ò": 45,
+    "ù": 37,
+    "û": 44,
+    "œ": 39,
+    "ɓ": 33,
+    "ɗ": 21,
+    "ə": 7,
+    "ʼ": 32,
+    "̀": 35,
+    "̂": 36
+  },
+  "mqf": {
+    "'": 46,
+    "-": 20,
+    "0": 30,
+    "1": 32,
+    "2": 33,
+    "3": 37,
+    "4": 31,
+    "5": 35,
+    "6": 38,
+    "7": 40,
+    "8": 42,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 11,
+    "c": 45,
+    "d": 43,
+    "e": 5,
+    "g": 47,
+    "h": 41,
+    "i": 14,
+    "j": 44,
+    "k": 13,
+    "l": 34,
+    "m": 12,
+    "n": 8,
+    "o": 6,
+    "p": 36,
+    "r": 10,
+    "s": 16,
+    "t": 9,
+    "u": 17,
+    "w": 18,
+    "y": 15,
+    "|": 4,
+    "à": 23,
+    "á": 19,
+    "è": 27,
+    "é": 25,
+    "ì": 28,
+    "í": 24,
+    "ò": 21,
+    "ó": 22,
+    "ù": 26,
+    "ú": 29
+  },
+  "mqj": {
+    "'": 17,
+    "-": 23,
+    "0": 36,
+    "1": 26,
+    "2": 27,
+    "3": 29,
+    "4": 30,
+    "5": 31,
+    "6": 32,
+    "7": 33,
+    "8": 34,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 38,
+    "d": 19,
+    "e": 12,
+    "f": 28,
+    "g": 16,
+    "h": 25,
+    "i": 9,
+    "j": 39,
+    "k": 13,
+    "l": 8,
+    "m": 10,
+    "n": 6,
+    "o": 11,
+    "p": 18,
+    "r": 20,
+    "s": 15,
+    "t": 14,
+    "u": 7,
+    "w": 24,
+    "y": 22,
+    "z": 35,
+    "|": 5
+  },
+  "mqn": {
+    "'": 20,
+    "-": 26,
+    "1": 30,
+    "3": 32,
+    "4": 31,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 25,
+    "d": 14,
+    "e": 9,
+    "f": 27,
+    "g": 21,
+    "h": 19,
+    "i": 7,
+    "j": 28,
+    "k": 12,
+    "l": 17,
+    "m": 11,
+    "n": 8,
+    "o": 6,
+    "p": 15,
+    "r": 16,
+    "s": 18,
+    "t": 13,
+    "u": 10,
+    "w": 22,
+    "y": 24,
+    "z": 29,
+    "|": 5
+  },
+  "mri": {
+    "!": 59,
+    "$": 52,
+    "%": 53,
+    "&": 55,
+    "'": 51,
+    "+": 56,
+    ",": 47,
+    "-": 31,
+    ".": 43,
+    "/": 48,
+    "0": 28,
+    "1": 29,
+    "2": 34,
+    "3": 40,
+    "4": 38,
+    "5": 37,
+    "6": 41,
+    "7": 45,
+    "8": 42,
+    "9": 36,
+    ":": 50,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 60,
+    "[": 57,
+    "]": 58,
+    "a": 5,
+    "b": 30,
+    "c": 25,
+    "d": 27,
+    "e": 7,
+    "f": 33,
+    "g": 16,
+    "h": 11,
+    "i": 6,
+    "j": 39,
+    "k": 9,
+    "l": 24,
+    "m": 17,
+    "n": 10,
+    "o": 12,
+    "p": 19,
+    "q": 49,
+    "r": 13,
+    "s": 22,
+    "t": 8,
+    "u": 14,
+    "v": 35,
+    "w": 18,
+    "x": 46,
+    "y": 32,
+    "z": 44,
+    "|": 4,
+    "ā": 15,
+    "ē": 21,
+    "ī": 26,
+    "ō": 20,
+    "ū": 23
+  },
+  "mrw": {
+    "'": 23,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "d": 19,
+    "e": 12,
+    "g": 11,
+    "h": 22,
+    "i": 8,
+    "k": 9,
+    "l": 18,
+    "m": 13,
+    "n": 6,
+    "o": 7,
+    "p": 16,
+    "r": 17,
+    "s": 10,
+    "t": 15,
+    "w": 21,
+    "y": 14,
+    "|": 5
+  },
+  "msy": {
+    "'": 33,
+    "-": 40,
+    "0": 27,
+    "1": 29,
+    "2": 30,
+    "3": 34,
+    "4": 32,
+    "5": 36,
+    "6": 37,
+    "7": 31,
+    "8": 39,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "d": 21,
+    "e": 12,
+    "f": 24,
+    "g": 7,
+    "h": 13,
+    "i": 6,
+    "j": 26,
+    "k": 16,
+    "l": 25,
+    "m": 8,
+    "n": 11,
+    "o": 18,
+    "p": 23,
+    "r": 10,
+    "s": 22,
+    "t": 19,
+    "u": 15,
+    "v": 20,
+    "w": 28,
+    "y": 35,
+    "z": 14,
+    "|": 5,
+    "ɨ": 9
+  },
+  "mtd": {
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 26,
+    "d": 13,
+    "e": 6,
+    "g": 16,
+    "h": 15,
+    "i": 8,
+    "j": 24,
+    "k": 12,
+    "l": 20,
+    "m": 14,
+    "n": 7,
+    "o": 22,
+    "p": 23,
+    "r": 17,
+    "s": 18,
+    "t": 10,
+    "u": 11,
+    "w": 25,
+    "y": 21,
+    "|": 4,
+    "ꞌ": 9
+  },
+  "mtj": {
+    "'": 27,
+    "-": 28,
+    "0": 30,
+    "1": 33,
+    "2": 34,
+    "4": 31,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 24,
+    "c": 25,
+    "d": 17,
+    "e": 7,
+    "f": 18,
+    "g": 11,
+    "h": 20,
+    "i": 8,
+    "j": 19,
+    "k": 15,
+    "l": 22,
+    "m": 14,
+    "n": 9,
+    "o": 5,
+    "p": 26,
+    "r": 10,
+    "s": 13,
+    "t": 16,
+    "u": 12,
+    "w": 23,
+    "y": 21,
+    "z": 29,
+    "|": 4
+  },
+  "mto": {
+    "'": 10,
+    "0": 40,
+    "1": 39,
+    "4": 42,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 34,
+    "c": 24,
+    "d": 22,
+    "e": 6,
+    "f": 35,
+    "g": 33,
+    "h": 26,
+    "i": 16,
+    "j": 8,
+    "k": 13,
+    "l": 30,
+    "m": 17,
+    "n": 12,
+    "o": 18,
+    "p": 20,
+    "q": 37,
+    "r": 25,
+    "s": 14,
+    "t": 9,
+    "u": 11,
+    "v": 19,
+    "w": 38,
+    "x": 21,
+    "y": 15,
+    "z": 36,
+    "|": 4,
+    "á": 23,
+    "é": 32,
+    "í": 29,
+    "ñ": 28,
+    "ó": 27,
+    "ú": 31,
+    "̱": 5
+  },
+  "muh": {
+    "'": 33,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 31,
+    "d": 16,
+    "e": 6,
+    "f": 21,
+    "g": 10,
+    "h": 26,
+    "i": 11,
+    "j": 32,
+    "k": 9,
+    "l": 18,
+    "m": 14,
+    "n": 8,
+    "o": 23,
+    "p": 22,
+    "r": 12,
+    "s": 25,
+    "t": 13,
+    "u": 20,
+    "v": 34,
+    "w": 17,
+    "y": 19,
+    "z": 28,
+    "|": 4,
+    "ï": 24,
+    "ö": 29,
+    "ü": 7,
+    "ʼ": 30
+  },
+  "mup": {
+    "'": 51,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 19,
+    "ः": 53,
+    "अ": 25,
+    "आ": 41,
+    "इ": 39,
+    "ई": 31,
+    "उ": 30,
+    "ऊ": 56,
+    "ए": 48,
+    "ऐ": 59,
+    "ओ": 58,
+    "क": 7,
+    "ख": 32,
+    "ग": 28,
+    "घ": 49,
+    "च": 35,
+    "छ": 47,
+    "ज": 22,
+    "झ": 52,
+    "ञ": 54,
+    "ट": 42,
+    "ठ": 50,
+    "ड": 33,
+    "ढ": 55,
+    "ण": 21,
+    "त": 16,
+    "थ": 38,
+    "द": 24,
+    "ध": 45,
+    "न": 12,
+    "प": 20,
+    "फ": 46,
+    "ब": 23,
+    "भ": 40,
+    "म": 11,
+    "य": 15,
+    "र": 8,
+    "ल": 29,
+    "ळ": 37,
+    "व": 26,
+    "ष": 57,
+    "स": 10,
+    "ह": 14,
+    "़": 36,
+    "ा": 5,
+    "ि": 27,
+    "ी": 13,
+    "ु": 18,
+    "ू": 34,
+    "े": 6,
+    "ो": 9,
+    "्": 17,
+    "‍": 44
+  },
+  "mur": {
+    "0": 34,
+    "1": 35,
+    "2": 36,
+    "4": 38,
+    "5": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 17,
+    "d": 24,
+    "e": 9,
+    "g": 16,
+    "i": 6,
+    "j": 26,
+    "k": 10,
+    "l": 13,
+    "m": 21,
+    "n": 7,
+    "o": 8,
+    "p": 29,
+    "r": 18,
+    "s": 27,
+    "t": 12,
+    "u": 14,
+    "v": 28,
+    "w": 25,
+    "y": 20,
+    "z": 19,
+    "|": 4,
+    "á": 31,
+    "ã": 30,
+    "ŋ": 22,
+    "ɔ": 11,
+    "ɛ": 15,
+    "ḏ": 32,
+    "ṯ": 33
+  },
+  "muv": {
+    "'": 45,
+    "0": 49,
+    "3": 48,
+    "4": 47,
+    "6": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ം": 19,
+    "അ": 28,
+    "ആ": 37,
+    "ഇ": 33,
+    "ഈ": 41,
+    "ഉ": 40,
+    "ഊ": 44,
+    "എ": 35,
+    "ഏ": 39,
+    "ഒ": 36,
+    "ഓ": 43,
+    "ക": 6,
+    "ങ": 21,
+    "ച": 27,
+    "ഞ": 29,
+    "ട": 11,
+    "ണ": 26,
+    "ത": 8,
+    "ന": 7,
+    "പ": 13,
+    "മ": 20,
+    "യ": 15,
+    "ര": 23,
+    "റ": 12,
+    "ല": 18,
+    "ള": 16,
+    "വ": 14,
+    "ശ": 24,
+    "ാ": 10,
+    "ി": 9,
+    "ീ": 31,
+    "ു": 22,
+    "ൂ": 38,
+    "െ": 17,
+    "േ": 25,
+    "ൊ": 30,
+    "ോ": 34,
+    "്": 4,
+    "ൻ": 32,
+    "ർ": 42
+  },
+  "muy": {
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 33,
+    "d": 20,
+    "e": 14,
+    "f": 26,
+    "g": 15,
+    "h": 18,
+    "i": 6,
+    "j": 32,
+    "k": 10,
+    "l": 13,
+    "m": 8,
+    "n": 7,
+    "o": 28,
+    "p": 34,
+    "r": 17,
+    "s": 22,
+    "t": 11,
+    "u": 16,
+    "v": 29,
+    "w": 27,
+    "y": 12,
+    "z": 25,
+    "|": 4,
+    "à": 23,
+    "á": 41,
+    "â": 40,
+    "è": 35,
+    "é": 48,
+    "ê": 46,
+    "ì": 36,
+    "í": 49,
+    "î": 43,
+    "ò": 42,
+    "ô": 45,
+    "ù": 44,
+    "ú": 50,
+    "ŋ": 21,
+    "ɓ": 38,
+    "ɗ": 24,
+    "ə": 9,
+    "ʉ": 30,
+    "̀": 31,
+    "́": 47,
+    "̂": 39
+  },
+  "mvp": {
+    "'": 18,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 26,
+    "d": 20,
+    "e": 16,
+    "f": 28,
+    "g": 14,
+    "h": 23,
+    "i": 7,
+    "j": 22,
+    "k": 13,
+    "l": 12,
+    "m": 11,
+    "n": 6,
+    "o": 8,
+    "p": 17,
+    "q": 30,
+    "r": 19,
+    "s": 15,
+    "t": 9,
+    "u": 10,
+    "w": 25,
+    "y": 27,
+    "z": 29,
+    "|": 5
+  },
+  "mwq": {
+    "'": 21,
+    "-": 31,
+    "0": 38,
+    "1": 44,
+    "4": 41,
+    "6": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 12,
+    "d": 27,
+    "e": 14,
+    "f": 32,
+    "g": 11,
+    "h": 8,
+    "i": 6,
+    "j": 30,
+    "k": 9,
+    "l": 20,
+    "m": 13,
+    "n": 7,
+    "o": 24,
+    "p": 17,
+    "q": 46,
+    "r": 29,
+    "s": 19,
+    "t": 15,
+    "u": 10,
+    "v": 25,
+    "w": 16,
+    "x": 33,
+    "y": 26,
+    "z": 28,
+    "|": 4,
+    "á": 37,
+    "ä": 18,
+    "ç": 43,
+    "è": 34,
+    "é": 35,
+    "ë": 36,
+    "î": 42,
+    "ô": 40,
+    "õ": 45,
+    "ü": 22
+  },
+  "mwv": {
+    "'": 25,
+    "-": 26,
+    "0": 31,
+    "1": 33,
+    "4": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 24,
+    "d": 20,
+    "e": 9,
+    "f": 27,
+    "g": 14,
+    "h": 23,
+    "i": 6,
+    "j": 22,
+    "k": 10,
+    "l": 15,
+    "m": 13,
+    "n": 8,
+    "o": 19,
+    "p": 17,
+    "r": 18,
+    "s": 12,
+    "t": 11,
+    "u": 7,
+    "v": 28,
+    "w": 29,
+    "y": 21,
+    "z": 30,
+    "|": 5
+  },
+  "mxb": {
+    "'": 11,
+    "-": 48,
+    "0": 46,
+    "1": 38,
+    "2": 39,
+    "3": 40,
+    "4": 41,
+    "5": 42,
+    "6": 43,
+    "7": 44,
+    "8": 45,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 33,
+    "c": 25,
+    "d": 10,
+    "e": 20,
+    "f": 35,
+    "g": 34,
+    "h": 26,
+    "i": 8,
+    "j": 31,
+    "k": 13,
+    "l": 30,
+    "m": 29,
+    "n": 5,
+    "o": 9,
+    "p": 32,
+    "q": 37,
+    "r": 24,
+    "s": 18,
+    "t": 14,
+    "u": 16,
+    "v": 27,
+    "x": 23,
+    "y": 22,
+    "z": 36,
+    "|": 4,
+    "á": 12,
+    "é": 28,
+    "í": 15,
+    "ñ": 19,
+    "ó": 17,
+    "ú": 21,
+    "ü": 49,
+    "̱": 7
+  },
+  "mxq": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 23,
+    "d": 15,
+    "e": 18,
+    "f": 36,
+    "g": 24,
+    "i": 12,
+    "j": 9,
+    "k": 13,
+    "l": 30,
+    "m": 14,
+    "n": 11,
+    "o": 19,
+    "p": 20,
+    "r": 27,
+    "s": 16,
+    "t": 8,
+    "u": 22,
+    "v": 37,
+    "w": 25,
+    "x": 21,
+    "y": 7,
+    "z": 38,
+    "|": 4,
+    "á": 28,
+    "é": 29,
+    "í": 32,
+    "ñ": 26,
+    "ó": 33,
+    "ú": 31,
+    "ʉ": 5,
+    "́": 34,
+    "̱": 17,
+    "—": 35
+  },
+  "mxt": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 6,
+    "d": 19,
+    "e": 20,
+    "f": 36,
+    "g": 31,
+    "h": 7,
+    "i": 8,
+    "j": 30,
+    "k": 38,
+    "l": 27,
+    "m": 23,
+    "n": 10,
+    "o": 12,
+    "p": 33,
+    "q": 25,
+    "r": 14,
+    "s": 18,
+    "t": 11,
+    "u": 9,
+    "v": 16,
+    "x": 28,
+    "y": 13,
+    "z": 37,
+    "|": 4,
+    "á": 24,
+    "é": 34,
+    "í": 22,
+    "ñ": 15,
+    "ó": 26,
+    "ú": 29,
+    "ɨ": 17,
+    "́": 35,
+    "̱": 21,
+    "ṉ": 39
+  },
+  "mxv": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 31,
+    "c": 21,
+    "d": 11,
+    "e": 20,
+    "f": 40,
+    "g": 39,
+    "h": 22,
+    "i": 6,
+    "j": 24,
+    "k": 9,
+    "l": 25,
+    "m": 23,
+    "n": 7,
+    "o": 12,
+    "p": 30,
+    "q": 44,
+    "r": 13,
+    "s": 18,
+    "t": 16,
+    "u": 8,
+    "v": 15,
+    "x": 14,
+    "y": 19,
+    "z": 43,
+    "|": 4,
+    "à": 36,
+    "á": 37,
+    "â": 49,
+    "ä": 27,
+    "è": 48,
+    "é": 34,
+    "ë": 38,
+    "ì": 45,
+    "í": 29,
+    "î": 47,
+    "ï": 32,
+    "ñ": 17,
+    "ò": 46,
+    "ó": 42,
+    "ö": 35,
+    "ù": 41,
+    "ú": 26,
+    "ü": 28,
+    "̱": 33,
+    "ꞌ": 10
+  },
+  "mya": {
+    "\"": 44,
+    "$": 116,
+    "%": 118,
+    "'": 108,
+    "+": 125,
+    ",": 102,
+    "-": 85,
+    ".": 89,
+    "/": 106,
+    "0": 50,
+    "1": 54,
+    "2": 63,
+    "3": 88,
+    "4": 83,
+    "5": 80,
+    "6": 86,
+    "7": 92,
+    "8": 82,
+    "9": 76,
+    ":": 111,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 129,
+    "]": 130,
+    "a": 48,
+    "b": 84,
+    "c": 71,
+    "d": 73,
+    "e": 51,
+    "f": 96,
+    "g": 78,
+    "h": 69,
+    "i": 59,
+    "j": 113,
+    "k": 81,
+    "l": 65,
+    "m": 75,
+    "n": 53,
+    "o": 56,
+    "p": 77,
+    "q": 115,
+    "r": 58,
+    "s": 60,
+    "t": 61,
+    "u": 74,
+    "v": 101,
+    "w": 94,
+    "x": 110,
+    "y": 90,
+    "z": 119,
+    "£": 127,
+    "°": 121,
+    "²": 133,
+    "á": 122,
+    "í": 132,
+    "ó": 135,
+    "õ": 124,
+    "ú": 134,
+    "ł": 126,
+    "က": 9,
+    "ခ": 26,
+    "ဂ": 43,
+    "ဃ": 138,
+    "င": 8,
+    "စ": 20,
+    "ဆ": 32,
+    "ဇ": 57,
+    "ဈ": 114,
+    "ဉ": 45,
+    "ည": 22,
+    "ဋ": 117,
+    "ဌ": 97,
+    "ဍ": 105,
+    "ဏ": 55,
+    "တ": 16,
+    "ထ": 35,
+    "ဒ": 42,
+    "���": 67,
+    "န": 15,
+    "ပ": 14,
+    "ဖ": 30,
+    "ဗ": 49,
+    "ဘ": 37,
+    "မ": 13,
+    "ယ": 34,
+    "ရ": 19,
+    "လ": 27,
+    "ဝ": 40,
+    "သ": 12,
+    "ဟ": 38,
+    "ဠ": 112,
+    "အ": 21,
+    "ဣ": 137,
+    "ဤ": 70,
+    "ဥ": 68,
+    "ဦ": 62,
+    "ဧ": 98,
+    "ဩ": 109,
+    "ဪ": 139,
+    "ါ": 31,
+    "ာ": 6,
+    "ိ": 10,
+    "ီ": 28,
+    "ု": 7,
+    "ူ": 36,
+    "ေ": 11,
+    "ဲ": 33,
+    "ံ": 29,
+    "့": 18,
+    "း": 5,
+    "္": 41,
+    "်": 4,
+    "ျ": 24,
+    "ြ": 17,
+    "ွ": 25,
+    "ှ": 23,
+    "ဿ": 104,
+    "၀": 64,
+    "၁": 72,
+    "၂": 87,
+    "၃": 95,
+    "၄": 93,
+    "၅": 91,
+    "၆": 99,
+    "၇": 107,
+    "၈": 103,
+    "၉": 100,
+    "၊": 66,
+    "။": 47,
+    "၌": 79,
+    "၍": 52,
+    "၎": 46,
+    "၏": 39,
+    "ၤ": 136,
+    "‌": 120,
+    "–": 131,
+    "—": 128,
+    "”": 123
+  },
+  "myb": {
+    "'": 23,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 10,
+    "e": 11,
+    "g": 13,
+    "h": 24,
+    "i": 15,
+    "j": 19,
+    "k": 7,
+    "l": 14,
+    "m": 16,
+    "n": 6,
+    "o": 12,
+    "p": 32,
+    "r": 17,
+    "s": 22,
+    "t": 9,
+    "u": 20,
+    "w": 27,
+    "y": 25,
+    "|": 4,
+    "à": 35,
+    "á": 36,
+    "ã": 31,
+    "é": 33,
+    "ë": 34,
+    "î": 29,
+    "ï": 8,
+    "ð": 26,
+    "ò": 38,
+    "ó": 40,
+    "õ": 30,
+    "ø": 18,
+    "ù": 42,
+    "ú": 41,
+    "ü": 37,
+    "þ": 28
+  },
+  "myk": {
+    "'": 33,
+    "-": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 36,
+    "a": 5,
+    "b": 18,
+    "c": 32,
+    "d": 26,
+    "e": 8,
+    "f": 27,
+    "g": 16,
+    "h": 23,
+    "i": 6,
+    "j": 31,
+    "k": 13,
+    "l": 17,
+    "m": 20,
+    "n": 7,
+    "o": 15,
+    "p": 14,
+    "r": 19,
+    "s": 21,
+    "t": 24,
+    "u": 10,
+    "v": 39,
+    "w": 11,
+    "x": 35,
+    "y": 12,
+    "z": 34,
+    "|": 4,
+    "à": 38,
+    "á": 40,
+    "è": 37,
+    "é": 45,
+    "ì": 43,
+    "í": 44,
+    "ò": 46,
+    "ù": 41,
+    "ŋ": 29,
+    "ɔ": 22,
+    "ɛ": 9,
+    "ɲ": 25,
+    "ʼ": 30,
+    "̀": 47,
+    "́": 48,
+    "ᴐ": 28
+  },
+  "myl": {
+    "'": 24,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 18,
+    "e": 9,
+    "f": 28,
+    "g": 19,
+    "h": 17,
+    "i": 6,
+    "j": 27,
+    "k": 16,
+    "l": 14,
+    "m": 11,
+    "n": 7,
+    "o": 10,
+    "p": 13,
+    "r": 15,
+    "s": 21,
+    "t": 12,
+    "u": 8,
+    "w": 22,
+    "y": 23,
+    "z": 29,
+    "|": 5,
+    "–": 30
+  },
+  "myv": {
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 6,
+    "б": 30,
+    "в": 13,
+    "г": 28,
+    "д": 18,
+    "е": 5,
+    "ж": 32,
+    "з": 20,
+    "и": 17,
+    "й": 26,
+    "к": 12,
+    "л": 15,
+    "м": 16,
+    "н": 10,
+    "о": 9,
+    "п": 22,
+    "р": 14,
+    "с": 7,
+    "т": 8,
+    "у": 23,
+    "ф": 37,
+    "х": 36,
+    "ц": 29,
+    "ч": 27,
+    "ш": 24,
+    "щ": 38,
+    "ы": 25,
+    "ь": 11,
+    "э": 21,
+    "ю": 34,
+    "я": 19,
+    "ё": 31,
+    "–": 39,
+    "—": 33
+  },
+  "myx": {
+    "'": 24,
+    "-": 30,
+    "0": 31,
+    "1": 34,
+    "4": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 32,
+    "d": 21,
+    "e": 8,
+    "f": 23,
+    "g": 22,
+    "h": 15,
+    "i": 6,
+    "j": 27,
+    "k": 11,
+    "l": 12,
+    "m": 17,
+    "n": 9,
+    "o": 13,
+    "p": 25,
+    "r": 20,
+    "s": 14,
+    "t": 19,
+    "u": 7,
+    "v": 29,
+    "w": 16,
+    "y": 18,
+    "z": 28,
+    "|": 4,
+    "ŋ": 26
+  },
+  "myy": {
+    "'": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 16,
+    "d": 24,
+    "e": 11,
+    "f": 40,
+    "g": 21,
+    "h": 41,
+    "i": 6,
+    "j": 10,
+    "k": 48,
+    "l": 30,
+    "m": 13,
+    "n": 17,
+    "o": 8,
+    "p": 29,
+    "q": 28,
+    "r": 7,
+    "s": 15,
+    "t": 12,
+    "u": 22,
+    "v": 43,
+    "w": 27,
+    "x": 47,
+    "y": 14,
+    "z": 46,
+    "|": 4,
+    "á": 38,
+    "ã": 20,
+    "é": 37,
+    "í": 36,
+    "ñ": 23,
+    "ó": 42,
+    "õ": 25,
+    "ú": 31,
+    "ü": 45,
+    "ĩ": 18,
+    "ũ": 33,
+    "ʉ": 9,
+    "́": 39,
+    "̃": 26,
+    "ẽ": 32,
+    "‍": 35,
+    "—": 34
+  },
+  "mza": {
+    "'": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 40,
+    "c": 15,
+    "d": 12,
+    "e": 21,
+    "f": 47,
+    "g": 29,
+    "h": 16,
+    "i": 7,
+    "j": 22,
+    "k": 11,
+    "l": 30,
+    "m": 18,
+    "n": 6,
+    "o": 13,
+    "p": 37,
+    "q": 51,
+    "r": 14,
+    "s": 24,
+    "t": 10,
+    "u": 8,
+    "v": 19,
+    "x": 33,
+    "y": 20,
+    "z": 49,
+    "|": 4,
+    "à": 23,
+    "á": 38,
+    "ä": 45,
+    "è": 44,
+    "é": 41,
+    "ë": 52,
+    "ì": 26,
+    "í": 35,
+    "ï": 43,
+    "ñ": 17,
+    "ò": 50,
+    "ó": 46,
+    "ö": 34,
+    "ù": 25,
+    "ú": 42,
+    "ü": 31,
+    "ā": 27,
+    "ē": 39,
+    "ī": 28,
+    "ō": 48,
+    "ū": 36,
+    "̱": 32
+  },
+  "mzi": {
+    "'": 15,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "c": 27,
+    "d": 21,
+    "e": 14,
+    "f": 37,
+    "g": 17,
+    "h": 28,
+    "i": 6,
+    "j": 9,
+    "k": 10,
+    "l": 32,
+    "m": 24,
+    "n": 5,
+    "o": 11,
+    "p": 35,
+    "r": 30,
+    "s": 13,
+    "t": 12,
+    "u": 22,
+    "x": 16,
+    "y": 25,
+    "|": 4,
+    "á": 19,
+    "é": 23,
+    "í": 18,
+    "ñ": 31,
+    "ó": 20,
+    "ú": 29,
+    "ń": 34,
+    "̱": 8,
+    "ṉ": 33,
+    "–": 36
+  },
+  "mzj": {
+    "'": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 20,
+    "e": 14,
+    "f": 22,
+    "g": 27,
+    "h": 30,
+    "i": 6,
+    "j": 24,
+    "k": 11,
+    "l": 8,
+    "m": 12,
+    "n": 7,
+    "o": 13,
+    "p": 33,
+    "s": 17,
+    "t": 16,
+    "u": 15,
+    "v": 25,
+    "w": 21,
+    "y": 18,
+    "z": 31,
+    "|": 4,
+    "à": 26,
+    "ǃ": 35,
+    "ǹ": 28,
+    "ɔ": 9,
+    "ɛ": 10,
+    "ɲ": 29,
+    "̀": 32,
+    "́": 34
+  },
+  "mzk": {
+    "'": 39,
+    "-": 25,
+    "1": 42,
+    "2": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 10,
+    "c": 29,
+    "d": 14,
+    "e": 6,
+    "f": 36,
+    "g": 19,
+    "h": 5,
+    "i": 11,
+    "j": 22,
+    "k": 20,
+    "l": 9,
+    "m": 17,
+    "n": 7,
+    "o": 12,
+    "p": 32,
+    "r": 23,
+    "s": 18,
+    "t": 21,
+    "u": 13,
+    "v": 33,
+    "w": 24,
+    "x": 41,
+    "y": 28,
+    "z": 40,
+    "|": 4,
+    "à": 16,
+    "á": 30,
+    "è": 27,
+    "é": 15,
+    "ì": 35,
+    "í": 37,
+    "ò": 26,
+    "ó": 34,
+    "ù": 38,
+    "ú": 31
+  },
+  "mzm": {
+    "'": 49,
+    "-": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "d": 24,
+    "e": 16,
+    "f": 38,
+    "g": 9,
+    "h": 26,
+    "i": 10,
+    "j": 31,
+    "k": 12,
+    "l": 20,
+    "m": 27,
+    "n": 6,
+    "o": 23,
+    "p": 21,
+    "r": 14,
+    "s": 17,
+    "t": 15,
+    "u": 11,
+    "v": 28,
+    "w": 22,
+    "y": 19,
+    "z": 18,
+    "|": 4,
+    "à": 35,
+    "á": 36,
+    "ã": 29,
+    "è": 40,
+    "é": 42,
+    "ì": 32,
+    "í": 39,
+    "ñ": 44,
+    "ò": 45,
+    "ó": 48,
+    "õ": 50,
+    "ù": 41,
+    "ú": 43,
+    "ĩ": 51,
+    "ń": 47,
+    "ň": 58,
+    "ŏ": 56,
+    "ũ": 37,
+    "ǎ": 46,
+    "ǐ": 54,
+    "ǹ": 52,
+    "ɓ": 13,
+    "ɔ": 7,
+    "ɛ": 8,
+    "̀": 33,
+    "́": 34,
+    "̃": 30,
+    "̆": 57,
+    "̌": 55,
+    "ṹ": 53
+  },
+  "mzw": {
+    "'": 31,
+    "-": 35,
+    "6": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 28,
+    "d": 14,
+    "e": 7,
+    "f": 32,
+    "g": 21,
+    "h": 18,
+    "i": 12,
+    "j": 29,
+    "k": 16,
+    "l": 11,
+    "m": 22,
+    "n": 9,
+    "o": 8,
+    "p": 25,
+    "r": 17,
+    "s": 20,
+    "t": 19,
+    "u": 27,
+    "v": 34,
+    "w": 13,
+    "y": 23,
+    "|": 4,
+    "á": 36,
+    "â": 6,
+    "å": 24,
+    "é": 37,
+    "ó": 39,
+    "ô": 15,
+    "ú": 42,
+    "ń": 40,
+    "ŋ": 33,
+    "ɔ": 30,
+    "ɛ": 26,
+    "́": 38
+  },
+  "nab": {
+    "'": 43,
+    "0": 35,
+    "1": 11,
+    "2": 6,
+    "3": 5,
+    "4": 36,
+    "5": 38,
+    "6": 40,
+    "7": 39,
+    "8": 41,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 32,
+    "e": 18,
+    "f": 34,
+    "h": 13,
+    "i": 9,
+    "j": 23,
+    "k": 15,
+    "l": 14,
+    "m": 31,
+    "n": 8,
+    "o": 25,
+    "p": 30,
+    "r": 29,
+    "s": 19,
+    "t": 12,
+    "u": 16,
+    "v": 33,
+    "w": 20,
+    "x": 7,
+    "y": 21,
+    "|": 10,
+    "ã": 17,
+    "õ": 37,
+    "ĩ": 26,
+    "ũ": 24,
+    "̱": 22,
+    "ẽ": 27,
+    "—": 28
+  },
+  "nag": {
+    "'": 30,
+    "-": 26,
+    "0": 31,
+    "1": 35,
+    "2": 38,
+    "3": 37,
+    "4": 33,
+    "5": 36,
+    "6": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 23,
+    "d": 22,
+    "e": 7,
+    "f": 29,
+    "g": 20,
+    "h": 10,
+    "i": 6,
+    "j": 21,
+    "k": 9,
+    "l": 17,
+    "m": 18,
+    "n": 11,
+    "o": 8,
+    "p": 19,
+    "q": 34,
+    "r": 13,
+    "s": 15,
+    "t": 12,
+    "u": 14,
+    "v": 27,
+    "w": 25,
+    "x": 32,
+    "y": 24,
+    "z": 28,
+    "|": 4
+  },
+  "nan": {
+    "'": 49,
+    "-": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 12,
+    "b": 32,
+    "c": 13,
+    "e": 19,
+    "g": 11,
+    "h": 5,
+    "i": 7,
+    "j": 42,
+    "k": 10,
+    "l": 15,
+    "m": 30,
+    "n": 8,
+    "o": 16,
+    "p": 23,
+    "s": 14,
+    "t": 9,
+    "u": 28,
+    "|": 4,
+    "à": 33,
+    "á": 27,
+    "â": 20,
+    "è": 34,
+    "é": 45,
+    "ê": 17,
+    "ì": 36,
+    "í": 25,
+    "î": 40,
+    "ò": 35,
+    "ó": 21,
+    "ô": 37,
+    "ù": 39,
+    "ú": 38,
+    "û": 43,
+    "ā": 26,
+    "ē": 41,
+    "ī": 29,
+    "ń": 48,
+    "ō": 22,
+    "ū": 24,
+    "ǹ": 46,
+    "̂": 47,
+    "̄": 44,
+    "̍": 18,
+    "͘": 31,
+    "ḿ": 50
+  },
+  "nas": {
+    "'": 10,
+    "-": 19,
+    "0": 25,
+    "1": 27,
+    "2": 30,
+    "3": 34,
+    "4": 29,
+    "5": 33,
+    "6": 28,
+    "7": 31,
+    "8": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "d": 18,
+    "e": 9,
+    "g": 15,
+    "h": 35,
+    "i": 8,
+    "j": 23,
+    "k": 11,
+    "l": 24,
+    "m": 16,
+    "n": 6,
+    "o": 7,
+    "p": 17,
+    "r": 14,
+    "s": 22,
+    "t": 13,
+    "u": 12,
+    "v": 20,
+    "|": 5,
+    "–": 26
+  },
+  "naw": {
+    "'": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 27,
+    "e": 23,
+    "f": 20,
+    "g": 22,
+    "h": 30,
+    "i": 12,
+    "k": 11,
+    "l": 18,
+    "m": 10,
+    "n": 7,
+    "o": 26,
+    "p": 25,
+    "r": 24,
+    "s": 16,
+    "t": 17,
+    "u": 19,
+    "w": 21,
+    "y": 15,
+    "|": 4,
+    "á": 32,
+    "ŋ": 29,
+    "ɔ": 14,
+    "ɛ": 8,
+    "ɩ": 6,
+    "ʋ": 9,
+    "́": 33,
+    "‐": 28
+  },
+  "nca": {
+    "'": 36,
+    "-": 28,
+    "0": 33,
+    "1": 35,
+    "2": 37,
+    "3": 42,
+    "4": 38,
+    "5": 40,
+    "6": 44,
+    "7": 45,
+    "8": 46,
+    "9": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "d": 19,
+    "e": 6,
+    "f": 34,
+    "g": 24,
+    "h": 25,
+    "i": 11,
+    "j": 31,
+    "k": 12,
+    "l": 30,
+    "m": 16,
+    "n": 8,
+    "o": 5,
+    "p": 21,
+    "q": 20,
+    "r": 10,
+    "s": 17,
+    "t": 15,
+    "u": 9,
+    "v": 39,
+    "w": 18,
+    "y": 14,
+    "z": 41,
+    "|": 4,
+    "á": 29,
+    "é": 27,
+    "í": 22,
+    "ó": 23,
+    "ú": 32,
+    "ŋ": 13
+  },
+  "nch": {
+    "'": 32,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "4": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 10,
+    "d": 25,
+    "e": 11,
+    "f": 31,
+    "g": 33,
+    "h": 13,
+    "i": 6,
+    "j": 17,
+    "k": 39,
+    "l": 16,
+    "m": 14,
+    "n": 9,
+    "o": 12,
+    "p": 19,
+    "q": 15,
+    "r": 21,
+    "s": 18,
+    "t": 7,
+    "u": 8,
+    "v": 35,
+    "x": 22,
+    "y": 20,
+    "z": 23,
+    "|": 5,
+    "á": 28,
+    "é": 30,
+    "í": 24,
+    "ó": 34,
+    "ú": 26,
+    "—": 29
+  },
+  "ncj": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 26,
+    "c": 13,
+    "d": 21,
+    "e": 12,
+    "f": 30,
+    "g": 27,
+    "h": 10,
+    "i": 5,
+    "j": 23,
+    "k": 34,
+    "l": 14,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "q": 15,
+    "r": 24,
+    "s": 17,
+    "t": 9,
+    "u": 8,
+    "v": 32,
+    "x": 20,
+    "y": 18,
+    "z": 22,
+    "|": 4,
+    "á": 29,
+    "é": 28,
+    "í": 31,
+    "ñ": 35,
+    "ó": 33,
+    "ú": 25
+  },
+  "ncl": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 29,
+    "c": 10,
+    "d": 23,
+    "e": 11,
+    "f": 33,
+    "g": 30,
+    "h": 14,
+    "i": 6,
+    "j": 19,
+    "k": 36,
+    "l": 9,
+    "m": 15,
+    "n": 7,
+    "o": 12,
+    "p": 18,
+    "q": 17,
+    "r": 25,
+    "s": 16,
+    "t": 13,
+    "u": 8,
+    "v": 32,
+    "x": 21,
+    "y": 20,
+    "z": 24,
+    "|": 4,
+    "á": 31,
+    "é": 28,
+    "í": 22,
+    "ñ": 37,
+    "ó": 34,
+    "ú": 26,
+    "—": 27,
+    "ꞌ": 35
+  },
+  "ncu": {
+    "'": 30,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 14,
+    "d": 28,
+    "e": 6,
+    "f": 24,
+    "g": 26,
+    "h": 31,
+    "i": 17,
+    "k": 13,
+    "l": 27,
+    "m": 11,
+    "n": 9,
+    "o": 8,
+    "p": 25,
+    "r": 16,
+    "s": 15,
+    "t": 23,
+    "u": 22,
+    "w": 20,
+    "y": 12,
+    "|": 4,
+    "ò": 29,
+    "ŋ": 19,
+    "ɔ": 10,
+    "ɛ": 18,
+    "ʻ": 32,
+    "ʼ": 33,
+    "̱": 5
+  },
+  "ndj": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 23,
+    "d": 19,
+    "e": 18,
+    "f": 25,
+    "g": 17,
+    "h": 15,
+    "i": 6,
+    "j": 26,
+    "k": 11,
+    "l": 8,
+    "m": 10,
+    "n": 9,
+    "o": 16,
+    "p": 22,
+    "r": 29,
+    "s": 20,
+    "t": 21,
+    "u": 7,
+    "v": 14,
+    "w": 12,
+    "y": 13,
+    "z": 28,
+    "|": 5
+  },
+  "ndp": {
+    "'": 16,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 15,
+    "d": 14,
+    "e": 8,
+    "f": 27,
+    "g": 22,
+    "h": 25,
+    "i": 5,
+    "k": 10,
+    "l": 19,
+    "m": 13,
+    "n": 9,
+    "o": 7,
+    "p": 24,
+    "r": 12,
+    "s": 20,
+    "t": 17,
+    "u": 11,
+    "v": 26,
+    "w": 23,
+    "y": 18,
+    "z": 21,
+    "|": 4
+  },
+  "ndv": {
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 34,
+    "d": 28,
+    "e": 6,
+    "f": 23,
+    "g": 31,
+    "h": 11,
+    "i": 7,
+    "j": 36,
+    "k": 15,
+    "l": 14,
+    "m": 21,
+    "n": 8,
+    "o": 9,
+    "p": 29,
+    "r": 16,
+    "s": 20,
+    "t": 12,
+    "u": 13,
+    "w": 19,
+    "y": 10,
+    "|": 4,
+    "é": 25,
+    "ë": 17,
+    "í": 18,
+    "ñ": 32,
+    "ú": 26,
+    "ŋ": 35,
+    "ƴ": 33,
+    "ɓ": 24,
+    "ɗ": 22,
+    "ˈ": 30
+  },
+  "ndy": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "d": 16,
+    "e": 7,
+    "f": 31,
+    "g": 12,
+    "h": 33,
+    "i": 8,
+    "j": 26,
+    "k": 11,
+    "l": 20,
+    "m": 13,
+    "n": 6,
+    "o": 10,
+    "p": 32,
+    "r": 19,
+    "s": 23,
+    "t": 17,
+    "u": 9,
+    "v": 18,
+    "w": 30,
+    "y": 15,
+    "z": 14,
+    "|": 4,
+    "ɓ": 22,
+    "ɔ": 21,
+    "ɗ": 24,
+    "ə": 27,
+    "̱": 29
+  },
+  "ndz": {
+    "'": 18,
+    "-": 37,
+    "0": 47,
+    "1": 48,
+    "2": 50,
+    "4": 49,
+    "6": 54,
+    "7": 52,
+    "8": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 8,
+    "c": 30,
+    "d": 7,
+    "e": 21,
+    "f": 46,
+    "g": 13,
+    "i": 16,
+    "j": 26,
+    "k": 15,
+    "l": 29,
+    "m": 12,
+    "n": 5,
+    "o": 19,
+    "p": 38,
+    "r": 43,
+    "s": 40,
+    "t": 9,
+    "u": 25,
+    "v": 35,
+    "w": 32,
+    "y": 23,
+    "z": 36,
+    "|": 4,
+    "à": 17,
+    "á": 14,
+    "â": 27,
+    "è": 28,
+    "é": 34,
+    "ê": 45,
+    "ì": 11,
+    "í": 6,
+    "î": 39,
+    "ò": 22,
+    "ó": 20,
+    "ô": 44,
+    "ù": 24,
+    "ú": 33,
+    "û": 41,
+    "ŋ": 42,
+    "ṛ": 31,
+    "—": 51
+  },
+  "neb": {
+    "'": 18,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 49,
+    "a": 8,
+    "b": 16,
+    "d": 35,
+    "e": 10,
+    "f": 42,
+    "g": 13,
+    "h": 29,
+    "i": 11,
+    "k": 12,
+    "l": 7,
+    "m": 28,
+    "n": 6,
+    "o": 19,
+    "p": 26,
+    "s": 33,
+    "t": 22,
+    "u": 30,
+    "v": 44,
+    "w": 14,
+    "y": 24,
+    "z": 34,
+    "|": 4,
+    "à": 5,
+    "á": 15,
+    "â": 45,
+    "è": 31,
+    "é": 23,
+    "ê": 46,
+    "ì": 43,
+    "í": 36,
+    "î": 47,
+    "ò": 32,
+    "ó": 27,
+    "ô": 48,
+    "ù": 37,
+    "ú": 39,
+    "ɔ": 20,
+    "ɛ": 9,
+    "ɩ": 38,
+    "ʋ": 40,
+    "̀": 17,
+    "́": 25,
+    "̂": 41
+  },
+  "new": {
+    "'": 59,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 38,
+    "ं": 9,
+    "ः": 16,
+    "अ": 37,
+    "आ": 44,
+    "इ": 35,
+    "ई": 57,
+    "उ": 47,
+    "ऊ": 58,
+    "ए": 50,
+    "ओ": 60,
+    "क": 18,
+    "ख": 28,
+    "ग": 21,
+    "घ": 54,
+    "ङ": 62,
+    "च": 33,
+    "छ": 31,
+    "ज": 24,
+    "झ": 42,
+    "ञ": 55,
+    "ट": 46,
+    "ठ": 56,
+    "ड": 48,
+    "ढ": 66,
+    "ण": 51,
+    "त": 15,
+    "थ": 26,
+    "द": 27,
+    "ध": 30,
+    "न": 12,
+    "प": 22,
+    "फ": 40,
+    "ब": 32,
+    "भ": 41,
+    "म": 14,
+    "य": 7,
+    "र": 25,
+    "ल": 20,
+    "व": 13,
+    "श": 36,
+    "ष": 43,
+    "स": 19,
+    "ह": 23,
+    "ा": 5,
+    "ि": 8,
+    "ी": 29,
+    "ु": 10,
+    "ू": 34,
+    "ृ": 53,
+    "े": 17,
+    "ै": 49,
+    "ो": 45,
+    "ौ": 52,
+    "्": 6,
+    "०": 61,
+    "१": 63,
+    "२": 64,
+    "६": 65,
+    "‍": 11
+  },
+  "nfa": {
+    "'": 18,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 27,
+    "d": 12,
+    "e": 6,
+    "f": 29,
+    "g": 21,
+    "h": 11,
+    "i": 9,
+    "j": 24,
+    "k": 16,
+    "l": 17,
+    "m": 14,
+    "n": 10,
+    "o": 19,
+    "p": 22,
+    "r": 13,
+    "s": 15,
+    "t": 20,
+    "u": 8,
+    "w": 28,
+    "y": 25,
+    "|": 4,
+    "è": 7
+  },
+  "nfr": {
+    "'": 30,
+    "-": 33,
+    "2": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 29,
+    "d": 28,
+    "e": 7,
+    "f": 26,
+    "g": 24,
+    "h": 25,
+    "i": 8,
+    "j": 27,
+    "k": 10,
+    "l": 15,
+    "m": 12,
+    "n": 6,
+    "o": 17,
+    "p": 16,
+    "r": 9,
+    "s": 22,
+    "t": 21,
+    "u": 11,
+    "v": 32,
+    "w": 20,
+    "y": 13,
+    "z": 31,
+    "|": 4,
+    "ŋ": 23,
+    "ɛ": 14,
+    "ↄ": 18
+  },
+  "nga": {
+    "'": 21,
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 12,
+    "e": 10,
+    "f": 29,
+    "g": 14,
+    "h": 26,
+    "i": 8,
+    "k": 16,
+    "l": 13,
+    "m": 17,
+    "n": 6,
+    "o": 15,
+    "p": 28,
+    "r": 30,
+    "s": 23,
+    "t": 19,
+    "u": 20,
+    "v": 32,
+    "w": 11,
+    "y": 27,
+    "z": 22,
+    "|": 4,
+    "à": 40,
+    "á": 37,
+    "ã": 24,
+    "è": 39,
+    "é": 36,
+    "ì": 42,
+    "í": 41,
+    "ò": 45,
+    "ó": 44,
+    "ù": 35,
+    "ú": 43,
+    "ĩ": 33,
+    "ũ": 38,
+    "ɔ": 9,
+    "ɛ": 7,
+    "̀": 34,
+    "́": 31,
+    "̃": 25
+  },
+  "ngl": {
+    "'": 24,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 22,
+    "e": 8,
+    "f": 23,
+    "h": 11,
+    "i": 7,
+    "k": 15,
+    "l": 14,
+    "m": 13,
+    "n": 9,
+    "o": 6,
+    "p": 20,
+    "r": 17,
+    "s": 19,
+    "t": 18,
+    "u": 10,
+    "v": 21,
+    "w": 12,
+    "y": 16,
+    "|": 5
+  },
+  "ngp": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 27,
+    "d": 18,
+    "e": 9,
+    "f": 26,
+    "g": 16,
+    "h": 17,
+    "i": 6,
+    "j": 28,
+    "k": 11,
+    "l": 14,
+    "m": 12,
+    "n": 7,
+    "o": 13,
+    "p": 25,
+    "r": 29,
+    "s": 20,
+    "t": 19,
+    "u": 8,
+    "v": 23,
+    "w": 10,
+    "y": 15,
+    "z": 21,
+    "|": 4
+  },
+  "ngu": {
+    "0": 35,
+    "1": 36,
+    "2": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 13,
+    "d": 23,
+    "e": 8,
+    "f": 32,
+    "g": 29,
+    "h": 15,
+    "i": 6,
+    "j": 12,
+    "k": 38,
+    "l": 14,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "q": 17,
+    "r": 22,
+    "s": 18,
+    "t": 10,
+    "u": 9,
+    "v": 33,
+    "x": 21,
+    "y": 20,
+    "z": 24,
+    "|": 4,
+    "á": 27,
+    "é": 31,
+    "í": 28,
+    "ñ": 34,
+    "ó": 30,
+    "ú": 26
+  },
+  "nhe": {
+    "'": 23,
+    "-": 26,
+    "0": 31,
+    "1": 32,
+    "2": 33,
+    "3": 36,
+    "4": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 12,
+    "d": 25,
+    "e": 10,
+    "f": 28,
+    "g": 29,
+    "h": 14,
+    "i": 6,
+    "j": 16,
+    "k": 34,
+    "l": 11,
+    "m": 17,
+    "n": 9,
+    "o": 13,
+    "p": 18,
+    "q": 15,
+    "r": 22,
+    "s": 19,
+    "t": 7,
+    "u": 8,
+    "v": 30,
+    "x": 21,
+    "y": 20,
+    "z": 24,
+    "|": 5
+  },
+  "nhi": {
+    "'": 35,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 28,
+    "c": 10,
+    "d": 22,
+    "e": 14,
+    "f": 32,
+    "g": 30,
+    "h": 12,
+    "i": 5,
+    "j": 24,
+    "k": 37,
+    "l": 13,
+    "m": 15,
+    "n": 7,
+    "o": 9,
+    "p": 18,
+    "q": 17,
+    "r": 23,
+    "s": 16,
+    "t": 8,
+    "u": 11,
+    "v": 31,
+    "x": 21,
+    "y": 19,
+    "z": 20,
+    "|": 4,
+    "­": 38,
+    "á": 34,
+    "é": 29,
+    "í": 26,
+    "ñ": 36,
+    "ó": 33,
+    "ú": 25,
+    "—": 27
+  },
+  "nhu": {
+    "'": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 9,
+    "c": 26,
+    "d": 24,
+    "e": 5,
+    "f": 21,
+    "g": 23,
+    "h": 34,
+    "i": 12,
+    "j": 25,
+    "k": 13,
+    "l": 17,
+    "m": 16,
+    "n": 7,
+    "o": 8,
+    "p": 41,
+    "s": 19,
+    "t": 20,
+    "u": 10,
+    "v": 18,
+    "w": 11,
+    "y": 15,
+    "|": 4,
+    "à": 44,
+    "á": 43,
+    "â": 46,
+    "è": 39,
+    "é": 42,
+    "ê": 47,
+    "ì": 36,
+    "í": 40,
+    "î": 37,
+    "ò": 38,
+    "ó": 28,
+    "ô": 31,
+    "ù": 30,
+    "ú": 27,
+    "û": 45,
+    "ŋ": 22,
+    "ɛ": 6,
+    "̀": 33,
+    "́": 35,
+    "̂": 32
+  },
+  "nhw": {
+    "'": 32,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 10,
+    "d": 25,
+    "e": 12,
+    "f": 31,
+    "g": 33,
+    "h": 14,
+    "i": 6,
+    "j": 19,
+    "k": 39,
+    "l": 11,
+    "m": 15,
+    "n": 9,
+    "o": 13,
+    "p": 18,
+    "q": 16,
+    "r": 21,
+    "s": 17,
+    "t": 7,
+    "u": 8,
+    "v": 35,
+    "x": 22,
+    "y": 20,
+    "z": 23,
+    "|": 5,
+    "á": 28,
+    "é": 30,
+    "í": 24,
+    "ó": 34,
+    "ú": 26,
+    "—": 29
+  },
+  "nhx": {
+    "'": 19,
+    "3": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 35,
+    "c": 23,
+    "d": 25,
+    "e": 8,
+    "f": 36,
+    "g": 22,
+    "h": 24,
+    "i": 5,
+    "j": 10,
+    "k": 11,
+    "l": 18,
+    "m": 14,
+    "n": 9,
+    "o": 12,
+    "p": 21,
+    "q": 39,
+    "r": 26,
+    "s": 17,
+    "t": 13,
+    "u": 27,
+    "v": 38,
+    "w": 16,
+    "x": 28,
+    "y": 15,
+    "z": 29,
+    "|": 4,
+    "á": 20,
+    "é": 32,
+    "í": 31,
+    "ñ": 40,
+    "ó": 30,
+    "ú": 33,
+    "̱": 7,
+    "‐": 34,
+    "—": 37
+  },
+  "nhy": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 18,
+    "d": 22,
+    "e": 10,
+    "f": 34,
+    "g": 30,
+    "h": 19,
+    "i": 6,
+    "j": 12,
+    "k": 7,
+    "l": 13,
+    "m": 14,
+    "n": 8,
+    "o": 11,
+    "p": 17,
+    "q": 36,
+    "r": 21,
+    "s": 15,
+    "t": 9,
+    "u": 23,
+    "v": 35,
+    "w": 16,
+    "x": 24,
+    "y": 20,
+    "z": 25,
+    "|": 4,
+    "á": 33,
+    "é": 28,
+    "í": 31,
+    "ñ": 32,
+    "ó": 26,
+    "ú": 29
+  },
+  "nia": {
+    "'": 19,
+    "-": 28,
+    "0": 29,
+    "1": 36,
+    "2": 32,
+    "4": 33,
+    "6": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 16,
+    "e": 12,
+    "f": 23,
+    "g": 20,
+    "h": 17,
+    "i": 6,
+    "j": 31,
+    "k": 24,
+    "l": 13,
+    "m": 11,
+    "n": 8,
+    "o": 9,
+    "p": 30,
+    "r": 15,
+    "s": 18,
+    "t": 21,
+    "u": 10,
+    "w": 22,
+    "y": 26,
+    "z": 25,
+    "|": 4,
+    "ö": 7,
+    "̃": 27
+  },
+  "nij": {
+    "'": 27,
+    "-": 25,
+    "0": 28,
+    "1": 33,
+    "2": 34,
+    "4": 30,
+    "5": 32,
+    "6": 35,
+    "7": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 22,
+    "e": 7,
+    "f": 36,
+    "g": 17,
+    "h": 10,
+    "i": 9,
+    "j": 21,
+    "k": 11,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 14,
+    "p": 19,
+    "r": 16,
+    "s": 18,
+    "t": 8,
+    "u": 12,
+    "w": 23,
+    "y": 24,
+    "z": 29,
+    "|": 5
+  },
+  "nim": {
+    "'": 28,
+    "-": 27,
+    "0": 31,
+    "2": 32,
+    "4": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "d": 23,
+    "e": 12,
+    "f": 29,
+    "g": 15,
+    "h": 25,
+    "i": 8,
+    "j": 26,
+    "k": 10,
+    "l": 11,
+    "m": 13,
+    "n": 7,
+    "o": 14,
+    "p": 20,
+    "r": 30,
+    "s": 17,
+    "t": 16,
+    "u": 6,
+    "w": 19,
+    "y": 21,
+    "z": 22,
+    "|": 5,
+    "ĩ": 9,
+    "ũ": 18
+  },
+  "nin": {
+    "'": 33,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 25,
+    "d": 20,
+    "e": 16,
+    "f": 34,
+    "g": 10,
+    "h": 18,
+    "i": 6,
+    "j": 28,
+    "k": 11,
+    "l": 26,
+    "m": 14,
+    "n": 7,
+    "o": 15,
+    "p": 23,
+    "r": 9,
+    "s": 13,
+    "t": 17,
+    "u": 8,
+    "v": 27,
+    "w": 21,
+    "y": 19,
+    "z": 22,
+    "|": 4,
+    "à": 24,
+    "á": 36,
+    "â": 32,
+    "è": 44,
+    "é": 48,
+    "ì": 35,
+    "í": 37,
+    "î": 47,
+    "ò": 45,
+    "ù": 29,
+    "ú": 43,
+    "û": 50,
+    "ā": 46,
+    "ě": 41,
+    "ī": 40,
+    "ū": 42,
+    "ǎ": 38,
+    "ɨ": 31,
+    "̀": 49,
+    "̱": 30
+  },
+  "nko": {
+    "'": 37,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 15,
+    "d": 28,
+    "e": 29,
+    "f": 25,
+    "g": 30,
+    "h": 26,
+    "i": 21,
+    "k": 18,
+    "l": 13,
+    "m": 10,
+    "n": 9,
+    "o": 24,
+    "p": 23,
+    "r": 32,
+    "s": 16,
+    "t": 17,
+    "u": 19,
+    "w": 22,
+    "y": 20,
+    "|": 4,
+    "á": 14,
+    "é": 34,
+    "í": 27,
+    "ó": 33,
+    "ú": 31,
+    "ŋ": 36,
+    "ɔ": 11,
+    "ɛ": 12,
+    "ɩ": 8,
+    "ʋ": 7,
+    "́": 5,
+    "—": 38
+  },
+  "nlc": {
+    "'": 20,
+    "-": 26,
+    "1": 31,
+    "2": 33,
+    "3": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 29,
+    "a": 5,
+    "b": 11,
+    "c": 30,
+    "d": 18,
+    "e": 7,
+    "f": 27,
+    "g": 16,
+    "h": 22,
+    "i": 14,
+    "j": 24,
+    "k": 8,
+    "l": 12,
+    "m": 13,
+    "n": 6,
+    "o": 9,
+    "p": 23,
+    "r": 19,
+    "s": 17,
+    "t": 25,
+    "u": 10,
+    "v": 32,
+    "w": 21,
+    "y": 15,
+    "z": 28,
+    "|": 4
+  },
+  "nld": {
+    "$": 64,
+    "%": 52,
+    "&": 54,
+    "'": 31,
+    "+": 65,
+    ",": 46,
+    "-": 32,
+    ".": 35,
+    "/": 49,
+    "0": 28,
+    "1": 30,
+    "2": 33,
+    "3": 40,
+    "4": 39,
+    "5": 38,
+    "6": 42,
+    "7": 44,
+    "8": 41,
+    "9": 37,
+    ":": 48,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 71,
+    "]": 72,
+    "a": 7,
+    "b": 22,
+    "c": 24,
+    "d": 12,
+    "e": 5,
+    "f": 27,
+    "g": 15,
+    "h": 17,
+    "i": 8,
+    "j": 25,
+    "k": 19,
+    "l": 14,
+    "m": 18,
+    "n": 6,
+    "o": 11,
+    "p": 21,
+    "q": 45,
+    "r": 10,
+    "s": 13,
+    "t": 9,
+    "u": 20,
+    "v": 16,
+    "w": 23,
+    "x": 36,
+    "y": 29,
+    "z": 26,
+    "|": 4,
+    "£": 67,
+    "°": 61,
+    "²": 58,
+    "à": 68,
+    "á": 51,
+    "â": 75,
+    "ã": 69,
+    "ä": 78,
+    "æ": 80,
+    "ç": 62,
+    "è": 66,
+    "é": 43,
+    "ê": 73,
+    "ë": 34,
+    "í": 56,
+    "î": 77,
+    "ï": 47,
+    "ó": 50,
+    "ô": 76,
+    "õ": 57,
+    "ö": 60,
+    "ù": 79,
+    "ú": 63,
+    "û": 74,
+    "ü": 55,
+    "ă": 85,
+    "ł": 70,
+    "š": 83,
+    "ţ": 84,
+    "́": 82,
+    "̇": 59,
+    "–": 81
+  },
+  "nlg": {
+    "'": 32,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 24,
+    "d": 21,
+    "e": 9,
+    "f": 29,
+    "g": 12,
+    "h": 13,
+    "i": 6,
+    "j": 23,
+    "k": 17,
+    "l": 15,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 22,
+    "q": 31,
+    "r": 14,
+    "s": 19,
+    "t": 10,
+    "u": 8,
+    "v": 18,
+    "w": 25,
+    "x": 30,
+    "y": 26,
+    "z": 27,
+    "|": 4
+  },
+  "nlk": {
+    "-": 24,
+    "0": 26,
+    "1": 30,
+    "2": 31,
+    "3": 32,
+    "4": 29,
+    "5": 33,
+    "6": 36,
+    "7": 37,
+    "8": 38,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 34,
+    "d": 20,
+    "e": 6,
+    "f": 22,
+    "g": 12,
+    "h": 10,
+    "i": 9,
+    "j": 27,
+    "k": 14,
+    "l": 13,
+    "m": 15,
+    "n": 7,
+    "o": 11,
+    "p": 23,
+    "r": 19,
+    "s": 16,
+    "t": 18,
+    "u": 8,
+    "w": 17,
+    "y": 25,
+    "z": 28,
+    "|": 4
+  },
+  "nmz": {
+    "'": 38,
+    "-": 27,
+    "1": 46,
+    "2": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 24,
+    "a": 5,
+    "b": 8,
+    "c": 44,
+    "d": 16,
+    "e": 9,
+    "f": 33,
+    "g": 13,
+    "h": 28,
+    "i": 7,
+    "j": 34,
+    "k": 15,
+    "l": 17,
+    "m": 10,
+    "n": 6,
+    "o": 22,
+    "p": 31,
+    "r": 18,
+    "s": 23,
+    "t": 11,
+    "u": 14,
+    "v": 32,
+    "w": 20,
+    "y": 26,
+    "|": 4,
+    "à": 25,
+    "á": 29,
+    "ì": 41,
+    "í": 42,
+    "ù": 39,
+    "ú": 40,
+    "ĥ": 36,
+    "ŋ": 30,
+    "ɔ": 21,
+    "ɛ": 19,
+    "ɦ": 12,
+    "́": 35,
+    "̈": 37,
+    "–": 43
+  },
+  "nnb": {
+    "-": 28,
+    "0": 31,
+    "1": 33,
+    "2": 36,
+    "3": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 9,
+    "c": 35,
+    "d": 25,
+    "e": 7,
+    "f": 29,
+    "g": 20,
+    "h": 24,
+    "i": 13,
+    "j": 34,
+    "k": 12,
+    "l": 19,
+    "m": 11,
+    "n": 6,
+    "o": 8,
+    "p": 27,
+    "r": 16,
+    "s": 18,
+    "t": 21,
+    "u": 14,
+    "v": 32,
+    "w": 17,
+    "y": 10,
+    "z": 26,
+    "|": 5,
+    "­": 30,
+    "ĩ": 15,
+    "ũ": 22,
+    "ʼ": 23
+  },
+  "nno": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 22,
+    "c": 31,
+    "d": 15,
+    "e": 5,
+    "f": 21,
+    "g": 11,
+    "h": 19,
+    "i": 9,
+    "j": 20,
+    "k": 13,
+    "l": 14,
+    "m": 17,
+    "n": 7,
+    "o": 16,
+    "p": 25,
+    "r": 6,
+    "s": 12,
+    "t": 10,
+    "u": 23,
+    "v": 18,
+    "y": 26,
+    "|": 4,
+    "å": 24,
+    "æ": 30,
+    "ò": 29,
+    "ø": 27
+  },
+  "nnq": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "c": 23,
+    "d": 20,
+    "e": 7,
+    "f": 26,
+    "g": 12,
+    "h": 21,
+    "i": 8,
+    "j": 18,
+    "k": 10,
+    "l": 15,
+    "m": 13,
+    "n": 6,
+    "o": 11,
+    "p": 19,
+    "s": 24,
+    "t": 17,
+    "u": 9,
+    "v": 28,
+    "w": 16,
+    "y": 22,
+    "z": 27,
+    "|": 5
+  },
+  "nnw": {
+    "'": 45,
+    "-": 38,
+    "0": 54,
+    "1": 49,
+    "2": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 32,
+    "b": 15,
+    "c": 28,
+    "d": 16,
+    "e": 34,
+    "f": 36,
+    "g": 46,
+    "h": 50,
+    "i": 18,
+    "j": 33,
+    "k": 17,
+    "l": 20,
+    "m": 19,
+    "n": 7,
+    "o": 26,
+    "p": 25,
+    "r": 22,
+    "s": 21,
+    "t": 11,
+    "u": 23,
+    "v": 41,
+    "w": 13,
+    "y": 14,
+    "z": 24,
+    "|": 4,
+    "à": 44,
+    "á": 39,
+    "è": 52,
+    "é": 53,
+    "ì": 31,
+    "í": 35,
+    "ò": 48,
+    "ó": 51,
+    "ù": 42,
+    "ú": 40,
+    "ý": 43,
+    "ń": 37,
+    "ŋ": 29,
+    "ɑ": 5,
+    "ɔ": 47,
+    "ə": 6,
+    "ɛ": 30,
+    "ɡ": 27,
+    "ɩ": 8,
+    "ʋ": 9,
+    "̀": 10,
+    "́": 12,
+    "̌": 55
+  },
+  "noa": {
+    "'": 14,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 17,
+    "d": 20,
+    "e": 12,
+    "f": 43,
+    "g": 16,
+    "h": 6,
+    "i": 7,
+    "j": 13,
+    "k": 19,
+    "l": 34,
+    "m": 8,
+    "n": 15,
+    "o": 24,
+    "p": 18,
+    "q": 49,
+    "r": 9,
+    "s": 26,
+    "t": 21,
+    "u": 10,
+    "v": 44,
+    "w": 22,
+    "x": 50,
+    "y": 36,
+    "z": 46,
+    "|": 5,
+    "á": 37,
+    "ã": 25,
+    "ä": 41,
+    "é": 42,
+    "ë": 27,
+    "í": 39,
+    "ñ": 47,
+    "ó": 45,
+    "õ": 32,
+    "ö": 28,
+    "ú": 38,
+    "ĩ": 29,
+    "ũ": 33,
+    "ʌ": 11,
+    "́": 48,
+    "̃": 35,
+    "̈": 30,
+    "ẽ": 31,
+    "—": 40
+  },
+  "nob": {
+    "!": 58,
+    "\"": 73,
+    "%": 54,
+    "&": 65,
+    "'": 53,
+    "+": 61,
+    ",": 51,
+    "-": 31,
+    ".": 34,
+    "/": 50,
+    "0": 28,
+    "1": 32,
+    "2": 33,
+    "3": 41,
+    "4": 40,
+    "5": 39,
+    "6": 42,
+    "7": 45,
+    "8": 43,
+    "9": 38,
+    ":": 47,
+    ";": 63,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 74,
+    "[": 70,
+    "]": 71,
+    "a": 10,
+    "b": 22,
+    "c": 29,
+    "d": 14,
+    "e": 5,
+    "f": 20,
+    "g": 16,
+    "h": 24,
+    "i": 11,
+    "j": 26,
+    "k": 15,
+    "l": 12,
+    "m": 17,
+    "n": 7,
+    "o": 13,
+    "p": 19,
+    "q": 49,
+    "r": 6,
+    "s": 9,
+    "t": 8,
+    "u": 21,
+    "v": 18,
+    "w": 35,
+    "x": 46,
+    "y": 25,
+    "z": 44,
+    "|": 4,
+    "¥": 64,
+    "«": 36,
+    "°": 67,
+    "²": 72,
+    "»": 37,
+    "á": 59,
+    "ã": 68,
+    "å": 23,
+    "æ": 30,
+    "ç": 60,
+    "é": 48,
+    "í": 56,
+    "ó": 69,
+    "õ": 57,
+    "ø": 27,
+    "ú": 62,
+    "ü": 55,
+    "ł": 66,
+    "–": 52
+  },
+  "nod": {
+    "'": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 16,
+    "ก": 11,
+    "ข": 28,
+    "ค": 26,
+    "ฆ": 60,
+    "ง": 12,
+    "จ": 20,
+    "ฉ": 61,
+    "ช": 58,
+    "ซ": 49,
+    "ญ": 48,
+    "ฎ": 64,
+    "ฏ": 63,
+    "ฐ": 59,
+    "ฑ": 66,
+    "ฒ": 65,
+    "ณ": 54,
+    "ด": 29,
+    "ต": 10,
+    "ถ": 47,
+    "ท": 50,
+    "ธ": 51,
+    "น": 6,
+    "บ": 23,
+    "ป": 27,
+    "ผ": 41,
+    "ฝ": 57,
+    "พ": 34,
+    "ฟ": 53,
+    "ภ": 56,
+    "ม": 15,
+    "ย": 13,
+    "ร": 21,
+    "ฤ": 62,
+    "ล": 32,
+    "ว": 19,
+    "ศ": 55,
+    "ษ": 52,
+    "ส": 33,
+    "ห": 17,
+    "ฬ": 68,
+    "อ": 7,
+    "ฮ": 38,
+    "ะ": 22,
+    "ั": 14,
+    "า": 4,
+    "ิ": 31,
+    "ี": 18,
+    "ึ": 43,
+    "ื": 30,
+    "ุ": 45,
+    "ู": 24,
+    "เ": 8,
+    "แ": 35,
+    "โ": 42,
+    "ใ": 37,
+    "ไ": 36,
+    "็": 40,
+    "่": 9,
+    "้": 5,
+    "๊": 44,
+    "๋": 25,
+    "์": 39,
+    "ํ": 46
+  },
+  "nog": {
+    "'": 42,
+    "-": 33,
+    "0": 36,
+    "1": 39,
+    "2": 41,
+    "4": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 19,
+    "в": 28,
+    "г": 18,
+    "д": 12,
+    "е": 8,
+    "ж": 34,
+    "з": 22,
+    "и": 7,
+    "й": 23,
+    "к": 13,
+    "л": 9,
+    "м": 16,
+    "н": 6,
+    "о": 17,
+    "п": 25,
+    "р": 11,
+    "с": 15,
+    "т": 14,
+    "у": 21,
+    "ф": 35,
+    "х": 30,
+    "ц": 40,
+    "ч": 37,
+    "ш": 26,
+    "ъ": 24,
+    "ы": 10,
+    "ь": 20,
+    "э": 27,
+    "ю": 32,
+    "я": 29,
+    "–": 31
+  },
+  "not": {
+    "'": 39,
+    "0": 29,
+    "1": 31,
+    "2": 35,
+    "3": 38,
+    "4": 36,
+    "5": 40,
+    "6": 37,
+    "7": 33,
+    "8": 41,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 15,
+    "d": 30,
+    "e": 12,
+    "g": 11,
+    "h": 25,
+    "i": 4,
+    "j": 21,
+    "l": 34,
+    "m": 16,
+    "n": 9,
+    "o": 7,
+    "p": 18,
+    "q": 17,
+    "r": 10,
+    "s": 13,
+    "t": 8,
+    "u": 14,
+    "v": 32,
+    "y": 22,
+    "z": 43,
+    "|": 6,
+    "á": 23,
+    "é": 28,
+    "ë": 19,
+    "í": 27,
+    "ó": 24,
+    "—": 26
+  },
+  "npi": {
+    "!": 120,
+    "$": 110,
+    "%": 108,
+    "'": 105,
+    "+": 114,
+    ",": 87,
+    "-": 61,
+    ".": 88,
+    "/": 100,
+    "0": 49,
+    "1": 51,
+    "2": 58,
+    "3": 70,
+    "4": 67,
+    "5": 63,
+    "6": 69,
+    "7": 80,
+    "8": 72,
+    "9": 62,
+    ":": 94,
+    ";": 107,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 111,
+    "a": 59,
+    "b": 89,
+    "c": 79,
+    "d": 91,
+    "e": 66,
+    "f": 97,
+    "g": 90,
+    "h": 82,
+    "i": 75,
+    "j": 98,
+    "k": 92,
+    "l": 77,
+    "m": 85,
+    "n": 68,
+    "o": 65,
+    "p": 83,
+    "q": 117,
+    "r": 73,
+    "s": 71,
+    "t": 81,
+    "u": 86,
+    "v": 99,
+    "w": 101,
+    "x": 102,
+    "y": 96,
+    "z": 106,
+    "|": 4,
+    "£": 119,
+    "¥": 112,
+    "°": 113,
+    "ã": 118,
+    "õ": 115,
+    "ँ": 45,
+    "ं": 48,
+    "ः": 76,
+    "अ": 32,
+    "आ": 44,
+    "इ": 47,
+    "ई": 41,
+    "उ": 35,
+    "ऊ": 103,
+    "ऋ": 109,
+    "ए": 31,
+    "ऐ": 104,
+    "ओ": 74,
+    "औ": 84,
+    "क": 9,
+    "ख": 46,
+    "ग": 20,
+    "घ": 52,
+    "ङ": 50,
+    "च": 38,
+    "छ": 29,
+    "ज": 28,
+    "झ": 60,
+    "ञ": 64,
+    "ट": 27,
+    "ठ": 54,
+    "ड": 39,
+    "ढ": 57,
+    "ण": 43,
+    "त": 16,
+    "थ": 34,
+    "द": 24,
+    "ध": 36,
+    "न": 8,
+    "प": 19,
+    "फ": 42,
+    "ब": 26,
+    "भ": 30,
+    "म": 14,
+    "य": 17,
+    "र": 7,
+    "ल": 15,
+    "व": 22,
+    "श": 37,
+    "ष": 40,
+    "स": 11,
+    "ह": 18,
+    "ा": 5,
+    "ि": 10,
+    "ी": 23,
+    "ु": 21,
+    "ू": 25,
+    "ृ": 56,
+    "े": 13,
+    "ै": 33,
+    "ो": 12,
+    "ौ": 55,
+    "्": 6,
+    "।": 53,
+    "१": 122,
+    "४": 123,
+    "९": 116,
+    "‌": 93,
+    "‍": 78,
+    "–": 121,
+    "”": 95
+  },
+  "npl": {
+    "'": 38,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 30,
+    "c": 18,
+    "d": 23,
+    "e": 11,
+    "f": 33,
+    "g": 32,
+    "h": 13,
+    "i": 6,
+    "j": 17,
+    "k": 7,
+    "l": 12,
+    "m": 14,
+    "n": 10,
+    "o": 8,
+    "p": 19,
+    "q": 36,
+    "r": 21,
+    "s": 15,
+    "t": 9,
+    "u": 22,
+    "v": 34,
+    "w": 16,
+    "x": 24,
+    "y": 20,
+    "z": 35,
+    "|": 4,
+    "á": 27,
+    "é": 29,
+    "í": 26,
+    "ñ": 37,
+    "ó": 25,
+    "ú": 28,
+    "—": 31
+  },
+  "npy": {
+    "'": 25,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 28,
+    "d": 18,
+    "e": 10,
+    "f": 27,
+    "g": 19,
+    "h": 14,
+    "i": 6,
+    "j": 26,
+    "k": 16,
+    "l": 12,
+    "m": 11,
+    "n": 7,
+    "o": 9,
+    "p": 13,
+    "r": 17,
+    "s": 21,
+    "t": 15,
+    "u": 8,
+    "w": 22,
+    "y": 23,
+    "|": 5
+  },
+  "nso": {
+    "!": 69,
+    "\"": 53,
+    "$": 48,
+    "%": 50,
+    "&": 58,
+    "'": 45,
+    "+": 52,
+    ",": 43,
+    "-": 32,
+    ".": 39,
+    "/": 47,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 36,
+    "4": 35,
+    "5": 34,
+    "6": 40,
+    "7": 41,
+    "8": 37,
+    "9": 33,
+    ":": 46,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 62,
+    "]": 63,
+    "a": 5,
+    "b": 14,
+    "c": 27,
+    "d": 18,
+    "e": 6,
+    "f": 25,
+    "g": 9,
+    "h": 19,
+    "i": 11,
+    "j": 26,
+    "k": 13,
+    "l": 8,
+    "m": 15,
+    "n": 12,
+    "o": 7,
+    "p": 23,
+    "q": 44,
+    "r": 20,
+    "s": 17,
+    "t": 10,
+    "u": 24,
+    "v": 31,
+    "w": 21,
+    "x": 42,
+    "y": 22,
+    "z": 38,
+    "|": 4,
+    "£": 61,
+    "¥": 59,
+    "°": 71,
+    "²": 68,
+    "á": 55,
+    "ã": 70,
+    "ç": 56,
+    "é": 66,
+    "í": 49,
+    "ó": 73,
+    "õ": 67,
+    "ú": 64,
+    "ü": 57,
+    "ł": 60,
+    "š": 16,
+    "̇": 74,
+    "–": 65,
+    "—": 51,
+    "”": 72
+  },
+  "nst": {
+    "'": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 24,
+    "d": 25,
+    "e": 12,
+    "g": 13,
+    "h": 11,
+    "i": 9,
+    "j": 26,
+    "k": 8,
+    "l": 21,
+    "m": 14,
+    "n": 6,
+    "o": 7,
+    "p": 19,
+    "r": 23,
+    "s": 16,
+    "t": 10,
+    "u": 17,
+    "v": 22,
+    "y": 15,
+    "z": 28,
+    "|": 4,
+    "ü": 27,
+    "ā": 20
+  },
+  "nsu": {
+    "0": 37,
+    "1": 39,
+    "2": 40,
+    "4": 41,
+    "5": 43,
+    "6": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 28,
+    "c": 19,
+    "d": 23,
+    "e": 9,
+    "f": 35,
+    "g": 30,
+    "h": 13,
+    "i": 5,
+    "j": 17,
+    "k": 7,
+    "l": 12,
+    "m": 14,
+    "n": 11,
+    "o": 10,
+    "p": 18,
+    "q": 36,
+    "r": 21,
+    "s": 15,
+    "t": 8,
+    "u": 22,
+    "v": 16,
+    "w": 26,
+    "x": 24,
+    "y": 20,
+    "z": 25,
+    "|": 4,
+    "á": 33,
+    "é": 31,
+    "í": 34,
+    "ñ": 38,
+    "ó": 27,
+    "ú": 29,
+    "—": 32
+  },
+  "ntm": {
+    "'": 36,
+    "-": 20,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 37,
+    "c": 35,
+    "d": 10,
+    "e": 22,
+    "f": 42,
+    "h": 23,
+    "i": 17,
+    "k": 5,
+    "m": 11,
+    "n": 6,
+    "o": 33,
+    "p": 12,
+    "s": 28,
+    "t": 15,
+    "u": 38,
+    "w": 30,
+    "y": 21,
+    "|": 4,
+    "à": 7,
+    "á": 13,
+    "è": 25,
+    "é": 27,
+    "ì": 24,
+    "í": 26,
+    "ò": 32,
+    "ó": 34,
+    "ù": 40,
+    "ú": 29,
+    "ń": 41,
+    "ǹ": 31,
+    "ɔ": 14,
+    "ɛ": 9,
+    "̀": 16,
+    "́": 18,
+    "̰": 19,
+    "ḭ": 39,
+    "ḿ": 44,
+    "ṵ": 43
+  },
+  "ntr": {
+    "'": 27,
+    "0": 30,
+    "3": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 19,
+    "e": 6,
+    "f": 26,
+    "g": 24,
+    "h": 28,
+    "i": 15,
+    "k": 12,
+    "l": 17,
+    "m": 16,
+    "n": 7,
+    "o": 9,
+    "p": 25,
+    "r": 23,
+    "s": 22,
+    "t": 20,
+    "u": 18,
+    "w": 21,
+    "y": 13,
+    "|": 4,
+    "ŋ": 10,
+    "ɛ": 8,
+    "ↄ": 11
+  },
+  "nuj": {
+    "'": 32,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 10,
+    "c": 30,
+    "d": 23,
+    "e": 7,
+    "f": 25,
+    "g": 15,
+    "h": 12,
+    "i": 6,
+    "j": 24,
+    "k": 26,
+    "l": 14,
+    "m": 13,
+    "n": 11,
+    "o": 8,
+    "p": 27,
+    "r": 17,
+    "s": 18,
+    "t": 20,
+    "u": 9,
+    "v": 29,
+    "w": 19,
+    "y": 16,
+    "z": 28,
+    "|": 5,
+    "ŋ": 21,
+    "ʼ": 22
+  },
+  "nus": {
+    "'": 35,
+    "-": 33,
+    "1": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 26,
+    "c": 14,
+    "d": 20,
+    "e": 21,
+    "g": 32,
+    "h": 19,
+    "i": 7,
+    "j": 23,
+    "k": 6,
+    "l": 16,
+    "m": 18,
+    "n": 9,
+    "o": 28,
+    "p": 31,
+    "r": 17,
+    "t": 10,
+    "u": 13,
+    "w": 30,
+    "y": 22,
+    "|": 4,
+    "ä": 12,
+    "ë": 27,
+    "ö": 24,
+    "ŋ": 25,
+    "ɔ": 15,
+    "ɛ": 5,
+    "ɣ": 29,
+    "̈": 34,
+    "̱": 11
+  },
+  "nuz": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 12,
+    "d": 22,
+    "e": 10,
+    "f": 30,
+    "g": 27,
+    "h": 15,
+    "i": 6,
+    "j": 13,
+    "k": 36,
+    "l": 14,
+    "m": 18,
+    "n": 7,
+    "o": 8,
+    "p": 19,
+    "q": 16,
+    "r": 21,
+    "s": 17,
+    "t": 11,
+    "u": 9,
+    "v": 35,
+    "x": 23,
+    "y": 20,
+    "z": 25,
+    "|": 4,
+    "á": 32,
+    "é": 29,
+    "í": 31,
+    "ñ": 34,
+    "ó": 33,
+    "ú": 28,
+    "—": 26
+  },
+  "nwb": {
+    "'": 5,
+    "-": 7,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 18,
+    "c": 33,
+    "d": 24,
+    "e": 9,
+    "f": 31,
+    "g": 22,
+    "h": 23,
+    "i": 13,
+    "j": 32,
+    "k": 16,
+    "l": 10,
+    "m": 19,
+    "n": 8,
+    "o": 20,
+    "p": 27,
+    "q": 35,
+    "r": 30,
+    "s": 29,
+    "t": 26,
+    "u": 21,
+    "v": 34,
+    "w": 28,
+    "y": 15,
+    "z": 25,
+    "|": 4,
+    "ɔ": 11,
+    "ɛ": 17,
+    "ɩ": 14,
+    "ʋ": 12
+  },
+  "nxq": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 13,
+    "c": 20,
+    "d": 15,
+    "e": 4,
+    "f": 24,
+    "g": 7,
+    "h": 11,
+    "i": 6,
+    "j": 22,
+    "k": 29,
+    "l": 8,
+    "m": 17,
+    "n": 12,
+    "o": 19,
+    "p": 28,
+    "q": 9,
+    "r": 26,
+    "s": 14,
+    "t": 23,
+    "u": 16,
+    "v": 18,
+    "w": 31,
+    "x": 30,
+    "y": 25,
+    "z": 21,
+    "|": 5
+  },
+  "nya": {
+    "!": 68,
+    "$": 49,
+    "%": 48,
+    "&": 57,
+    "'": 29,
+    "+": 61,
+    ",": 43,
+    "-": 42,
+    ".": 39,
+    "/": 46,
+    "0": 30,
+    "1": 31,
+    "2": 32,
+    "3": 37,
+    "4": 35,
+    "5": 34,
+    "6": 36,
+    "7": 40,
+    "8": 38,
+    "9": 33,
+    ":": 45,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 63,
+    "]": 64,
+    "a": 4,
+    "b": 25,
+    "c": 22,
+    "d": 16,
+    "e": 11,
+    "f": 26,
+    "g": 24,
+    "h": 15,
+    "i": 6,
+    "j": 28,
+    "k": 9,
+    "l": 14,
+    "m": 10,
+    "n": 7,
+    "o": 8,
+    "p": 20,
+    "q": 44,
+    "r": 21,
+    "s": 19,
+    "t": 13,
+    "u": 12,
+    "v": 27,
+    "w": 17,
+    "x": 41,
+    "y": 23,
+    "z": 18,
+    "|": 5,
+    "¥": 58,
+    "²": 69,
+    "á": 54,
+    "ã": 65,
+    "ç": 62,
+    "é": 59,
+    "í": 55,
+    "ó": 70,
+    "õ": 56,
+    "ö": 72,
+    "ú": 66,
+    "ü": 52,
+    "ŵ": 67,
+    "ʼ": 73,
+    "‘": 50,
+    "’": 51,
+    "”": 47,
+    "⁰": 71,
+    "◦": 60
+  },
+  "nyf": {
+    "'": 24,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 28,
+    "d": 18,
+    "e": 11,
+    "f": 26,
+    "g": 19,
+    "h": 12,
+    "i": 6,
+    "j": 27,
+    "k": 9,
+    "l": 16,
+    "m": 10,
+    "n": 8,
+    "o": 13,
+    "p": 29,
+    "r": 15,
+    "s": 21,
+    "t": 22,
+    "u": 7,
+    "v": 25,
+    "w": 14,
+    "y": 20,
+    "z": 17,
+    "|": 5,
+    "̱": 30
+  },
+  "nyn": {
+    "'": 21,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 13,
+    "c": 28,
+    "d": 23,
+    "e": 7,
+    "f": 27,
+    "g": 17,
+    "h": 18,
+    "i": 6,
+    "j": 24,
+    "k": 12,
+    "l": 25,
+    "m": 14,
+    "n": 10,
+    "o": 11,
+    "p": 26,
+    "r": 9,
+    "s": 20,
+    "t": 19,
+    "u": 8,
+    "v": 30,
+    "w": 15,
+    "y": 16,
+    "z": 22,
+    "|": 5
+  },
+  "nyo": {
+    "'": 23,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 11,
+    "c": 28,
+    "d": 24,
+    "e": 8,
+    "f": 26,
+    "g": 16,
+    "h": 19,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 22,
+    "m": 14,
+    "n": 9,
+    "o": 10,
+    "p": 27,
+    "r": 13,
+    "s": 20,
+    "t": 18,
+    "u": 7,
+    "v": 30,
+    "w": 15,
+    "y": 17,
+    "z": 21,
+    "|": 5
+  },
+  "nyy": {
+    "'": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "d": 27,
+    "e": 15,
+    "f": 26,
+    "g": 16,
+    "h": 32,
+    "i": 12,
+    "j": 23,
+    "k": 9,
+    "l": 8,
+    "m": 13,
+    "n": 7,
+    "o": 14,
+    "p": 22,
+    "r": 38,
+    "s": 17,
+    "t": 19,
+    "u": 6,
+    "v": 43,
+    "w": 18,
+    "y": 20,
+    "|": 4,
+    "ā": 24,
+    "ē": 31,
+    "ī": 30,
+    "ō": 28,
+    "ū": 29,
+    "ȳ": 46,
+    "̄": 25,
+    "̣": 39,
+    "ḅ": 44,
+    "ḳ": 35,
+    "ḷ": 42,
+    "ṃ": 41,
+    "ṣ": 40,
+    "ẉ": 47,
+    "ạ": 34,
+    "ẹ": 36,
+    "ị": 10,
+    "ọ": 37,
+    "ụ": 21,
+    "ỵ": 45
+  },
+  "nzi": {
+    "'": 29,
+    "-": 30,
+    "1": 31,
+    "2": 32,
+    "5": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "d": 19,
+    "e": 6,
+    "f": 28,
+    "g": 23,
+    "h": 22,
+    "i": 14,
+    "k": 16,
+    "l": 9,
+    "m": 11,
+    "n": 8,
+    "o": 15,
+    "p": 27,
+    "r": 26,
+    "s": 20,
+    "t": 24,
+    "u": 17,
+    "v": 25,
+    "w": 18,
+    "y": 13,
+    "z": 21,
+    "|": 4,
+    "ɔ": 10,
+    "ɛ": 7
+  },
+  "obo": {
+    "'": 26,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 23,
+    "c": 28,
+    "d": 8,
+    "e": 21,
+    "f": 29,
+    "g": 16,
+    "h": 24,
+    "i": 9,
+    "j": 27,
+    "k": 10,
+    "l": 20,
+    "m": 14,
+    "n": 6,
+    "o": 5,
+    "p": 17,
+    "q": 30,
+    "r": 19,
+    "s": 13,
+    "t": 11,
+    "u": 12,
+    "v": 25,
+    "w": 18,
+    "x": 32,
+    "y": 15,
+    "z": 31,
+    "|": 4
+  },
+  "oci": {
+    "!": 62,
+    "\"": 59,
+    "$": 68,
+    "%": 67,
+    "&": 71,
+    "'": 25,
+    "+": 77,
+    ",": 55,
+    "-": 48,
+    ".": 37,
+    "/": 58,
+    "0": 32,
+    "1": 36,
+    "2": 41,
+    "3": 47,
+    "4": 46,
+    "5": 45,
+    "6": 49,
+    "7": 52,
+    "8": 50,
+    "9": 44,
+    ":": 53,
+    ";": 57,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 61,
+    "[": 74,
+    "]": 75,
+    "a": 5,
+    "b": 21,
+    "c": 16,
+    "d": 14,
+    "e": 6,
+    "f": 23,
+    "g": 20,
+    "h": 27,
+    "i": 10,
+    "j": 29,
+    "k": 38,
+    "l": 13,
+    "m": 18,
+    "n": 8,
+    "o": 12,
+    "p": 17,
+    "q": 24,
+    "r": 9,
+    "s": 7,
+    "t": 11,
+    "u": 15,
+    "v": 22,
+    "w": 43,
+    "x": 33,
+    "y": 42,
+    "z": 30,
+    "|": 4,
+    "¥": 70,
+    "°": 73,
+    "²": 72,
+    "à": 35,
+    "á": 34,
+    "ã": 76,
+    "ç": 31,
+    "è": 19,
+    "é": 39,
+    "ë": 78,
+    "í": 40,
+    "ï": 51,
+    "ò": 26,
+    "ó": 54,
+    "õ": 69,
+    "ö": 80,
+    "ú": 65,
+    "ü": 56,
+    "ł": 79,
+    "–": 60,
+    "—": 63,
+    "‘": 64,
+    "’": 28,
+    "”": 66
+  },
+  "ojb-script_latin": {
+    "'": 17,
+    "-": 20,
+    "0": 25,
+    "2": 24,
+    "3": 26,
+    "5": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "c": 16,
+    "e": 13,
+    "h": 9,
+    "i": 4,
+    "k": 6,
+    "l": 22,
+    "m": 15,
+    "n": 8,
+    "o": 10,
+    "p": 18,
+    "r": 21,
+    "s": 12,
+    "t": 11,
+    "w": 14,
+    "y": 19,
+    "z": 27,
+    "|": 7,
+    "–": 23
+  },
+  "ojb-script_syllabics": {
+    "0": 125,
+    "1": 129,
+    "2": 123,
+    "3": 127,
+    "4": 133,
+    "5": 147,
+    "6": 151,
+    "7": 134,
+    "8": 145,
+    "9": 150,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 76,
+    "b": 122,
+    "c": 113,
+    "d": 110,
+    "e": 78,
+    "f": 146,
+    "g": 126,
+    "h": 101,
+    "i": 90,
+    "j": 109,
+    "k": 143,
+    "l": 96,
+    "m": 100,
+    "n": 107,
+    "o": 102,
+    "p": 104,
+    "q": 157,
+    "r": 91,
+    "s": 82,
+    "t": 105,
+    "u": 103,
+    "v": 130,
+    "w": 148,
+    "x": 80,
+    "y": 124,
+    "z": 135,
+    "|": 4,
+    "ᐁ": 23,
+    "ᐃ": 6,
+    "ᐄ": 79,
+    "ᐅ": 12,
+    "ᐆ": 86,
+    "ᐊ": 16,
+    "ᐋ": 41,
+    "ᐌ": 33,
+    "ᐎ": 21,
+    "ᐐ": 32,
+    "ᐗ": 37,
+    "ᐙ": 20,
+    "ᐞ": 55,
+    "ᐤ": 106,
+    "ᐧ": 155,
+    "ᐯ": 49,
+    "ᐱ": 30,
+    "ᐲ": 51,
+    "ᐳ": 88,
+    "ᐴ": 83,
+    "ᐸ": 38,
+    "ᐹ": 67,
+    "ᐺ": 70,
+    "ᐼ": 154,
+    "ᐾ": 137,
+    "ᑆ": 99,
+    "ᑉ": 118,
+    "ᑌ": 43,
+    "ᑎ": 29,
+    "ᑏ": 97,
+    "ᑐ": 62,
+    "ᑑ": 40,
+    "ᑕ": 8,
+    "ᑖ": 42,
+    "ᑗ": 120,
+    "ᑣ": 84,
+    "ᑦ": 89,
+    "ᑫ": 24,
+    "ᑭ": 9,
+    "ᑮ": 13,
+    "ᑯ": 25,
+    "ᑰ": 48,
+    "ᑲ": 14,
+    "ᑳ": 11,
+    "ᑴ": 63,
+    "ᑶ": 132,
+    "ᑸ": 108,
+    "ᑾ": 54,
+    "ᒀ": 58,
+    "ᒁ": 156,
+    "ᒃ": 17,
+    "ᒉ": 112,
+    "ᒋ": 10,
+    "ᒌ": 65,
+    "ᒍ": 144,
+    "ᒎ": 87,
+    "ᒐ": 98,
+    "ᒑ": 73,
+    "ᒒ": 158,
+    "ᒞ": 117,
+    "ᒡ": 15,
+    "ᒣ": 69,
+    "ᒥ": 28,
+    "ᒦ": 39,
+    "ᒧ": 56,
+    "ᒨ": 72,
+    "ᒪ": 31,
+    "ᒫ": 26,
+    "ᒬ": 131,
+    "ᒶ": 141,
+    "ᒸ": 140,
+    "ᒻ": 61,
+    "ᓀ": 50,
+    "ᓂ": 5,
+    "ᓃ": 46,
+    "ᓄ": 52,
+    "ᓅ": 60,
+    "ᓇ": 22,
+    "ᓈ": 34,
+    "ᓉ": 94,
+    "ᓋ": 138,
+    "ᓍ": 92,
+    "ᓐ": 7,
+    "ᓬ": 75,
+    "ᓭ": 68,
+    "ᓯ": 27,
+    "ᓰ": 57,
+    "ᓱ": 66,
+    "ᓲ": 93,
+    "ᓴ": 44,
+    "ᓵ": 71,
+    "ᓶ": 128,
+    "ᔀ": 149,
+    "ᔂ": 114,
+    "ᔅ": 64,
+    "ᔐ": 53,
+    "ᔑ": 19,
+    "ᔒ": 85,
+    "ᔓ": 81,
+    "ᔔ": 95,
+    "ᔕ": 59,
+    "ᔖ": 74,
+    "ᔗ": 119,
+    "ᔛ": 142,
+    "ᔡ": 153,
+    "ᔣ": 111,
+    "ᔥ": 18,
+    "ᔦ": 36,
+    "ᔨ": 116,
+    "ᔩ": 121,
+    "ᔪ": 136,
+    "ᔫ": 115,
+    "ᔭ": 47,
+    "ᔮ": 45,
+    "ᔾ": 152,
+    "ᕒ": 77,
+    "ᕽ": 35,
+    "–": 139
+  },
+  "oku": {
+    "-": 34,
+    "1": 50,
+    "3": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 14,
+    "b": 17,
+    "c": 28,
+    "d": 26,
+    "e": 5,
+    "f": 23,
+    "g": 12,
+    "h": 15,
+    "i": 9,
+    "j": 25,
+    "k": 11,
+    "l": 13,
+    "m": 19,
+    "n": 6,
+    "o": 20,
+    "r": 40,
+    "s": 10,
+    "t": 22,
+    "u": 18,
+    "v": 33,
+    "w": 21,
+    "y": 8,
+    "z": 43,
+    "|": 4,
+    "à": 31,
+    "á": 32,
+    "â": 47,
+    "è": 24,
+    "é": 37,
+    "ê": 48,
+    "ì": 41,
+    "í": 44,
+    "î": 39,
+    "ò": 35,
+    "ó": 38,
+    "ô": 42,
+    "ù": 49,
+    "ú": 45,
+    "û": 46,
+    "ŋ": 27,
+    "ə": 16,
+    "ɛ": 7,
+    "̀": 30,
+    "́": 29,
+    "̂": 36
+  },
+  "old": {
+    "'": 29,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 22,
+    "d": 17,
+    "e": 11,
+    "f": 19,
+    "g": 23,
+    "h": 15,
+    "i": 6,
+    "j": 28,
+    "k": 10,
+    "l": 16,
+    "m": 14,
+    "n": 7,
+    "o": 8,
+    "p": 26,
+    "r": 21,
+    "s": 18,
+    "t": 20,
+    "u": 9,
+    "v": 31,
+    "w": 13,
+    "y": 12,
+    "z": 30,
+    "|": 4,
+    "̱": 32,
+    "ḏ": 33,
+    "ḵ": 34,
+    "ṟ": 25
+  },
+  "omw": {
+    "-": 33,
+    "0": 22,
+    "1": 23,
+    "2": 24,
+    "3": 29,
+    "4": 26,
+    "5": 28,
+    "6": 27,
+    "7": 25,
+    "8": 31,
+    "9": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "e": 15,
+    "f": 32,
+    "h": 19,
+    "i": 6,
+    "k": 16,
+    "m": 12,
+    "n": 9,
+    "o": 14,
+    "p": 21,
+    "q": 13,
+    "r": 7,
+    "s": 20,
+    "t": 8,
+    "u": 11,
+    "v": 10,
+    "w": 18,
+    "y": 17,
+    "|": 5
+  },
+  "onb": {
+    "'": 27,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 12,
+    "b": 24,
+    "d": 18,
+    "e": 5,
+    "f": 6,
+    "g": 11,
+    "h": 7,
+    "i": 8,
+    "k": 20,
+    "l": 15,
+    "m": 16,
+    "n": 10,
+    "o": 9,
+    "p": 25,
+    "q": 26,
+    "s": 19,
+    "t": 17,
+    "u": 13,
+    "v": 14,
+    "x": 28,
+    "y": 21,
+    "z": 22,
+    "|": 4,
+    "—": 29
+  },
+  "ood": {
+    "'": 23,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 18,
+    "c": 17,
+    "d": 24,
+    "e": 9,
+    "f": 30,
+    "g": 12,
+    "h": 5,
+    "i": 7,
+    "j": 19,
+    "k": 14,
+    "l": 25,
+    "m": 8,
+    "n": 20,
+    "o": 11,
+    "p": 21,
+    "q": 32,
+    "r": 26,
+    "s": 13,
+    "t": 10,
+    "u": 16,
+    "v": 28,
+    "w": 15,
+    "x": 31,
+    "y": 27,
+    "z": 29,
+    "|": 4
+  },
+  "orm": {
+    "!": 57,
+    "\"": 53,
+    "$": 51,
+    "%": 52,
+    "&": 64,
+    "'": 27,
+    ",": 44,
+    "-": 35,
+    ".": 43,
+    "/": 48,
+    "0": 30,
+    "1": 32,
+    "2": 36,
+    "3": 42,
+    "4": 40,
+    "5": 38,
+    "6": 41,
+    "7": 45,
+    "8": 39,
+    "9": 37,
+    ":": 46,
+    ";": 59,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "=": 55,
+    "[": 61,
+    "]": 62,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 15,
+    "e": 8,
+    "f": 20,
+    "g": 21,
+    "h": 17,
+    "i": 6,
+    "j": 24,
+    "k": 14,
+    "l": 19,
+    "m": 16,
+    "n": 7,
+    "o": 11,
+    "p": 28,
+    "q": 25,
+    "r": 12,
+    "s": 13,
+    "t": 10,
+    "u": 9,
+    "v": 34,
+    "w": 23,
+    "x": 29,
+    "y": 22,
+    "z": 33,
+    "|": 5,
+    "£": 56,
+    "°": 60,
+    "í": 54,
+    "ó": 63,
+    "õ": 58,
+    "—": 50,
+    "‘": 49,
+    "’": 31,
+    "”": 47
+  },
+  "ory": {
+    "!": 128,
+    "\"": 107,
+    "$": 130,
+    "%": 104,
+    "'": 82,
+    "*": 121,
+    "+": 119,
+    ",": 89,
+    "-": 63,
+    ".": 73,
+    "/": 85,
+    "0": 54,
+    "1": 55,
+    "2": 61,
+    "3": 70,
+    "4": 68,
+    "5": 66,
+    "6": 69,
+    "7": 76,
+    "8": 71,
+    "9": 64,
+    ":": 91,
+    ";": 114,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 112,
+    "]": 113,
+    "a": 74,
+    "b": 92,
+    "c": 79,
+    "d": 88,
+    "e": 81,
+    "f": 106,
+    "g": 103,
+    "h": 95,
+    "i": 80,
+    "j": 123,
+    "k": 102,
+    "l": 94,
+    "m": 77,
+    "n": 78,
+    "o": 87,
+    "p": 83,
+    "q": 129,
+    "r": 90,
+    "s": 86,
+    "t": 84,
+    "u": 96,
+    "v": 108,
+    "w": 99,
+    "x": 111,
+    "y": 110,
+    "z": 109,
+    "|": 4,
+    "£": 127,
+    "¥": 115,
+    "­": 133,
+    "°": 125,
+    "õ": 120,
+    "।": 56,
+    "ଁ": 45,
+    "ଂ": 43,
+    "ଃ": 67,
+    "ଅ": 30,
+    "ଆ": 34,
+    "ଇ": 26,
+    "ଈ": 105,
+    "ଉ": 42,
+    "ଊ": 132,
+    "ଋ": 98,
+    "ଏ": 21,
+    "ଐ": 101,
+    "ଓ": 53,
+    "ଔ": 100,
+    "କ": 9,
+    "ଖ": 46,
+    "ଗ": 28,
+    "ଘ": 57,
+    "ଙ": 40,
+    "ଚ": 41,
+    "ଛ": 44,
+    "ଜ": 32,
+    "ଝ": 75,
+    "ଞ": 58,
+    "ଟ": 27,
+    "ଠ": 51,
+    "ଡ": 36,
+    "ଢ": 72,
+    "ଣ": 25,
+    "ତ": 13,
+    "ଥ": 22,
+    "ଦ": 23,
+    "ଧ": 37,
+    "ନ": 12,
+    "ପ": 15,
+    "ଫ": 48,
+    "ବ": 11,
+    "ଭ": 38,
+    "ମ": 17,
+    "ଯ": 29,
+    "ର": 7,
+    "ଲ": 20,
+    "ଳ": 39,
+    "ଵ": 62,
+    "ଶ": 33,
+    "ଷ": 35,
+    "ସ": 14,
+    "ହ": 18,
+    "଼": 50,
+    "ା": 6,
+    "ି": 8,
+    "ୀ": 31,
+    "ୁ": 16,
+    "ୂ": 47,
+    "ୃ": 52,
+    "ୄ": 124,
+    "େ": 10,
+    "ୈ": 65,
+    "ୋ": 24,
+    "ୌ": 59,
+    "୍": 5,
+    "ୗ": 122,
+    "ୟ": 19,
+    "୦": 118,
+    "୭": 126,
+    "ୱ": 49,
+    "‌": 60,
+    "‍": 97,
+    "–": 131,
+    "‘": 116,
+    "’": 93,
+    "”": 117
+  },
+  "oss": {
+    "'": 33,
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 7,
+    "б": 26,
+    "в": 28,
+    "г": 18,
+    "д": 8,
+    "е": 27,
+    "ж": 31,
+    "з": 19,
+    "и": 17,
+    "й": 15,
+    "к": 22,
+    "л": 23,
+    "м": 13,
+    "н": 10,
+    "о": 20,
+    "п": 29,
+    "р": 12,
+    "с": 14,
+    "т": 11,
+    "у": 9,
+    "ф": 24,
+    "х": 16,
+    "ц": 21,
+    "ч": 30,
+    "ъ": 25,
+    "ы": 6,
+    "э": 35,
+    "ю": 36,
+    "ӕ": 5,
+    "–": 32
+  },
+  "ote": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 27,
+    "d": 13,
+    "e": 10,
+    "f": 33,
+    "g": 15,
+    "h": 11,
+    "i": 7,
+    "j": 22,
+    "k": 42,
+    "l": 36,
+    "m": 18,
+    "n": 6,
+    "o": 16,
+    "p": 24,
+    "q": 35,
+    "r": 14,
+    "s": 23,
+    "t": 12,
+    "u": 9,
+    "v": 39,
+    "x": 26,
+    "y": 21,
+    "z": 34,
+    "|": 4,
+    "á": 25,
+    "é": 40,
+    "í": 32,
+    "ñ": 28,
+    "ó": 38,
+    "ø": 31,
+    "ú": 30,
+    "ü": 41,
+    "ɛ": 29,
+    "ʉ": 19,
+    "́": 37,
+    "̱": 17
+  },
+  "otq": {
+    "'": 11,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 39,
+    "d": 19,
+    "e": 12,
+    "f": 30,
+    "g": 16,
+    "h": 8,
+    "i": 6,
+    "j": 24,
+    "k": 27,
+    "l": 32,
+    "m": 14,
+    "n": 7,
+    "o": 15,
+    "p": 26,
+    "r": 17,
+    "s": 22,
+    "t": 13,
+    "u": 10,
+    "w": 29,
+    "x": 23,
+    "y": 21,
+    "z": 31,
+    "|": 4,
+    "á": 28,
+    "é": 37,
+    "ë": 34,
+    "í": 38,
+    "ñ": 25,
+    "ó": 36,
+    "ö": 18,
+    "ú": 33,
+    "̱": 9,
+    "—": 35
+  },
+  "ozm": {
+    "'": 55,
+    "-": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 16,
+    "b": 5,
+    "c": 42,
+    "d": 33,
+    "e": 9,
+    "f": 56,
+    "g": 34,
+    "h": 37,
+    "i": 31,
+    "j": 32,
+    "k": 30,
+    "l": 13,
+    "m": 7,
+    "n": 8,
+    "o": 26,
+    "p": 35,
+    "q": 60,
+    "r": 28,
+    "s": 19,
+    "t": 23,
+    "u": 39,
+    "v": 57,
+    "w": 24,
+    "x": 61,
+    "y": 11,
+    "z": 36,
+    "|": 4,
+    "á": 14,
+    "â": 38,
+    "é": 12,
+    "ê": 46,
+    "ë": 62,
+    "í": 21,
+    "î": 43,
+    "ï": 58,
+    "ó": 15,
+    "ô": 48,
+    "ø": 49,
+    "ú": 40,
+    "û": 47,
+    "ě": 41,
+    "ŋ": 44,
+    "œ": 54,
+    "ǎ": 45,
+    "ǐ": 51,
+    "ǒ": 52,
+    "ǔ": 59,
+    "ǿ": 50,
+    "ɑ": 27,
+    "ɔ": 25,
+    "ɛ": 17,
+    "ɨ": 10,
+    "ʉ": 18,
+    "ʼ": 20,
+    "́": 6,
+    "̂": 22,
+    "̌": 29,
+    "̍": 63
+  },
+  "pab": {
+    "'": 52,
+    "-": 31,
+    "0": 33,
+    "1": 38,
+    "2": 40,
+    "3": 48,
+    "4": 42,
+    "5": 44,
+    "6": 47,
+    "7": 43,
+    "8": 49,
+    "9": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 25,
+    "d": 24,
+    "e": 6,
+    "f": 26,
+    "g": 32,
+    "h": 9,
+    "i": 7,
+    "j": 20,
+    "k": 11,
+    "l": 18,
+    "m": 17,
+    "n": 12,
+    "o": 8,
+    "p": 27,
+    "q": 36,
+    "r": 14,
+    "s": 16,
+    "t": 10,
+    "u": 21,
+    "v": 35,
+    "w": 19,
+    "x": 15,
+    "y": 13,
+    "z": 39,
+    "|": 5,
+    "á": 29,
+    "â": 51,
+    "ã": 30,
+    "é": 28,
+    "ê": 45,
+    "í": 37,
+    "ó": 34,
+    "ô": 41,
+    "ú": 46,
+    "—": 22
+  },
+  "pad": {
+    "'": 14,
+    "-": 22,
+    "0": 39,
+    "1": 38,
+    "2": 41,
+    "3": 49,
+    "4": 42,
+    "5": 46,
+    "6": 47,
+    "7": 40,
+    "8": 50,
+    "9": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 26,
+    "d": 13,
+    "e": 19,
+    "f": 25,
+    "g": 23,
+    "h": 9,
+    "i": 5,
+    "j": 15,
+    "k": 10,
+    "l": 28,
+    "m": 18,
+    "n": 7,
+    "o": 8,
+    "p": 24,
+    "q": 34,
+    "r": 11,
+    "s": 17,
+    "t": 21,
+    "u": 20,
+    "v": 12,
+    "x": 31,
+    "z": 35,
+    "|": 6,
+    "á": 32,
+    "â": 43,
+    "ã": 30,
+    "ç": 48,
+    "é": 29,
+    "ê": 44,
+    "í": 36,
+    "î": 52,
+    "ó": 33,
+    "ô": 37,
+    "õ": 53,
+    "ú": 45,
+    "—": 27
+  },
+  "pag": {
+    "'": 25,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 24,
+    "d": 17,
+    "e": 12,
+    "f": 28,
+    "g": 14,
+    "h": 26,
+    "i": 7,
+    "j": 23,
+    "k": 13,
+    "l": 15,
+    "m": 16,
+    "n": 6,
+    "o": 8,
+    "p": 18,
+    "q": 32,
+    "r": 19,
+    "s": 10,
+    "t": 9,
+    "u": 21,
+    "v": 31,
+    "w": 22,
+    "x": 33,
+    "y": 11,
+    "z": 29,
+    "|": 5,
+    "–": 34,
+    "—": 30
+  },
+  "pam": {
+    "'": 30,
+    "-": 26,
+    "6": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 25,
+    "d": 19,
+    "e": 14,
+    "f": 27,
+    "g": 8,
+    "h": 24,
+    "i": 7,
+    "j": 23,
+    "k": 10,
+    "l": 12,
+    "m": 13,
+    "n": 6,
+    "o": 21,
+    "p": 16,
+    "q": 31,
+    "r": 20,
+    "s": 15,
+    "t": 11,
+    "u": 9,
+    "v": 29,
+    "w": 22,
+    "x": 32,
+    "y": 18,
+    "z": 28,
+    "|": 5,
+    "—": 33
+  },
+  "pan": {
+    "!": 81,
+    "\"": 76,
+    "%": 83,
+    "'": 55,
+    ",": 73,
+    "-": 53,
+    ".": 69,
+    "/": 79,
+    "0": 50,
+    "1": 54,
+    "2": 61,
+    "3": 71,
+    "4": 66,
+    "5": 65,
+    "6": 70,
+    "7": 72,
+    "8": 68,
+    "9": 63,
+    ":": 75,
+    ";": 77,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 80,
+    "c": 92,
+    "f": 93,
+    "i": 90,
+    "l": 91,
+    "x": 89,
+    "|": 4,
+    "¤": 94,
+    "¥": 88,
+    "°": 87,
+    "।": 36,
+    "ਂ": 16,
+    "ਅ": 28,
+    "ਆ": 22,
+    "ਇ": 35,
+    "ਈ": 39,
+    "ਉ": 38,
+    "ਊ": 64,
+    "ਏ": 51,
+    "ਐ": 58,
+    "ਓ": 60,
+    "ਔ": 78,
+    "ਕ": 11,
+    "ਖ": 41,
+    "ਗ": 31,
+    "ਘ": 52,
+    "ਚ": 29,
+    "ਛ": 57,
+    "ਜ": 21,
+    "ਝ": 56,
+    "ਟ": 34,
+    "ਠ": 62,
+    "ਡ": 42,
+    "ਢ": 67,
+    "ਣ": 37,
+    "ਤ": 15,
+    "ਥ": 46,
+    "ਦ": 13,
+    "ਧ": 43,
+    "ਨ": 12,
+    "ਪ": 24,
+    "ਫ": 44,
+    "ਬ": 32,
+    "ਭ": 45,
+    "ਮ": 23,
+    "ਯ": 49,
+    "ਰ": 6,
+    "ਲ": 17,
+    "ਵ": 18,
+    "ਸ": 8,
+    "ਹ": 14,
+    "਼": 26,
+    "ਾ": 5,
+    "ਿ": 9,
+    "ੀ": 7,
+    "ੁ": 30,
+    "ੂ": 33,
+    "ੇ": 10,
+    "ੈ": 27,
+    "ੋ": 20,
+    "ੌ": 48,
+    "੍": 40,
+    "ੜ": 47,
+    "ੰ": 25,
+    "ੱ": 19,
+    "​": 82,
+    "–": 84,
+    "—": 86,
+    "‘": 59,
+    "’": 85,
+    "”": 74
+  },
+  "pao": {
+    "'": 36,
+    "-": 23,
+    "1": 33,
+    "2": 34,
+    "3": 37,
+    "7": 35,
+    "8": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 28,
+    "d": 20,
+    "e": 8,
+    "f": 30,
+    "g": 21,
+    "h": 22,
+    "i": 17,
+    "j": 25,
+    "k": 12,
+    "l": 27,
+    "m": 10,
+    "n": 9,
+    "o": 6,
+    "p": 16,
+    "q": 32,
+    "r": 26,
+    "s": 13,
+    "t": 11,
+    "u": 7,
+    "v": 29,
+    "w": 14,
+    "x": 31,
+    "y": 19,
+    "z": 24,
+    "|": 4,
+    "ꞌ": 15
+  },
+  "pap": {
+    "'": 28,
+    "-": 37,
+    "0": 39,
+    "1": 44,
+    "2": 42,
+    "3": 50,
+    "4": 45,
+    "5": 52,
+    "6": 48,
+    "7": 49,
+    "8": 53,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 32,
+    "d": 14,
+    "e": 8,
+    "f": 23,
+    "g": 21,
+    "h": 20,
+    "i": 6,
+    "j": 35,
+    "k": 15,
+    "l": 19,
+    "m": 17,
+    "n": 7,
+    "o": 9,
+    "p": 18,
+    "q": 51,
+    "r": 12,
+    "s": 10,
+    "t": 11,
+    "u": 13,
+    "v": 34,
+    "w": 31,
+    "x": 46,
+    "y": 27,
+    "z": 36,
+    "|": 4,
+    "á": 22,
+    "è": 30,
+    "é": 24,
+    "í": 25,
+    "ñ": 26,
+    "ò": 29,
+    "ó": 33,
+    "ù": 40,
+    "ú": 38,
+    "ü": 41,
+    "—": 43
+  },
+  "pau": {
+    "'": 28,
+    "-": 32,
+    "0": 25,
+    "1": 33,
+    "2": 31,
+    "3": 37,
+    "4": 30,
+    "5": 34,
+    "6": 39,
+    "7": 38,
+    "8": 41,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 21,
+    "c": 20,
+    "d": 16,
+    "e": 5,
+    "f": 24,
+    "g": 15,
+    "h": 19,
+    "i": 8,
+    "j": 22,
+    "k": 11,
+    "l": 6,
+    "m": 10,
+    "n": 14,
+    "o": 12,
+    "p": 23,
+    "q": 42,
+    "r": 9,
+    "s": 17,
+    "t": 18,
+    "u": 13,
+    "v": 27,
+    "w": 36,
+    "x": 29,
+    "y": 26,
+    "|": 4,
+    "—": 40
+  },
+  "pbb": {
+    "'": 6,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 35,
+    "c": 13,
+    "d": 23,
+    "e": 7,
+    "f": 32,
+    "g": 28,
+    "h": 19,
+    "i": 14,
+    "j": 11,
+    "k": 43,
+    "l": 26,
+    "m": 20,
+    "n": 15,
+    "o": 29,
+    "p": 18,
+    "q": 31,
+    "r": 17,
+    "s": 9,
+    "t": 12,
+    "u": 10,
+    "v": 22,
+    "w": 16,
+    "x": 41,
+    "y": 8,
+    "z": 33,
+    "|": 4,
+    "á": 30,
+    "ã": 21,
+    "é": 34,
+    "í": 36,
+    "ñ": 38,
+    "ó": 40,
+    "ú": 42,
+    "ĩ": 27,
+    "ũ": 25,
+    "ẽ": 24,
+    "—": 37
+  },
+  "pbc": {
+    "'": 21,
+    "-": 44,
+    "0": 28,
+    "1": 30,
+    "2": 34,
+    "3": 38,
+    "4": 36,
+    "5": 37,
+    "6": 39,
+    "7": 33,
+    "8": 41,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 23,
+    "d": 26,
+    "e": 9,
+    "f": 35,
+    "g": 29,
+    "h": 22,
+    "i": 13,
+    "j": 24,
+    "k": 6,
+    "l": 14,
+    "m": 17,
+    "n": 8,
+    "o": 10,
+    "p": 7,
+    "q": 43,
+    "r": 25,
+    "s": 19,
+    "t": 11,
+    "u": 16,
+    "v": 32,
+    "w": 20,
+    "x": 40,
+    "y": 18,
+    "z": 31,
+    "|": 4,
+    "à": 12,
+    "ù": 15
+  },
+  "pbi": {
+    "-": 36,
+    "1": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 35,
+    "d": 9,
+    "e": 13,
+    "f": 30,
+    "g": 23,
+    "h": 22,
+    "i": 14,
+    "j": 34,
+    "k": 7,
+    "l": 11,
+    "m": 10,
+    "n": 8,
+    "p": 33,
+    "r": 19,
+    "s": 15,
+    "t": 12,
+    "u": 20,
+    "v": 26,
+    "w": 16,
+    "y": 25,
+    "z": 21,
+    "|": 4,
+    "à": 29,
+    "á": 28,
+    "ŋ": 17,
+    "ɓ": 31,
+    "ɗ": 24,
+    "ə": 6,
+    "ɨ": 27,
+    "́": 32,
+    "̌": 37
+  },
+  "pce": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ก": 24,
+    "ค": 32,
+    "ฆ": 25,
+    "ง": 22,
+    "จ": 37,
+    "ช": 39,
+    "ซ": 26,
+    "ฌ": 31,
+    "ณ": 33,
+    "ด": 9,
+    "ต": 14,
+    "ท": 38,
+    "น": 10,
+    "บ": 13,
+    "ป": 27,
+    "พ": 35,
+    "ฟ": 41,
+    "ม": 11,
+    "ย": 18,
+    "ร": 23,
+    "ล": 29,
+    "ว": 16,
+    "ห": 12,
+    "อ": 5,
+    "ฮ": 28,
+    "ะ": 30,
+    "ั": 20,
+    "า": 6,
+    "ี": 7,
+    "ึ": 19,
+    "ื": 36,
+    "ู": 21,
+    "เ": 8,
+    "แ": 15,
+    "โ": 17,
+    "็": 34
+  },
+  "pcm": {
+    "'": 32,
+    "-": 28,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "3": 38,
+    "4": 36,
+    "5": 35,
+    "7": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 23,
+    "c": 26,
+    "d": 10,
+    "e": 5,
+    "f": 22,
+    "g": 20,
+    "h": 24,
+    "i": 8,
+    "j": 27,
+    "k": 18,
+    "l": 15,
+    "m": 16,
+    "n": 9,
+    "o": 6,
+    "p": 21,
+    "r": 13,
+    "s": 11,
+    "t": 12,
+    "u": 19,
+    "v": 25,
+    "w": 17,
+    "x": 30,
+    "y": 14,
+    "z": 29,
+    "|": 4
+  },
+  "peg": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଂ": 19,
+    "ଃ": 40,
+    "ଅ": 38,
+    "ଆ": 21,
+    "ଇ": 20,
+    "ଉ": 37,
+    "ଏ": 34,
+    "ଓ": 39,
+    "କ": 12,
+    "ଗ": 17,
+    "ଙ": 30,
+    "ଚ": 27,
+    "ଜ": 14,
+    "ଞ": 36,
+    "ଟ": 33,
+    "ଡ": 29,
+    "ଣ": 32,
+    "ତ": 11,
+    "ଦ": 23,
+    "ନ": 10,
+    "ପ": 24,
+    "ବ": 25,
+    "ମ": 18,
+    "ର": 9,
+    "ଲ": 28,
+    "ସ": 22,
+    "ହ": 13,
+    "଼": 26,
+    "ା": 5,
+    "ି": 7,
+    "ୁ": 16,
+    "େ": 8,
+    "୍": 6,
+    "ୟ": 31,
+    "ୱ": 15,
+    "‍": 35
+  },
+  "pez": {
+    "'": 9,
+    "-": 28,
+    "0": 30,
+    "1": 33,
+    "2": 34,
+    "4": 31,
+    "6": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 32,
+    "d": 25,
+    "e": 8,
+    "g": 17,
+    "h": 7,
+    "i": 13,
+    "j": 23,
+    "k": 12,
+    "l": 15,
+    "m": 19,
+    "n": 6,
+    "o": 16,
+    "p": 18,
+    "r": 20,
+    "s": 21,
+    "t": 14,
+    "u": 10,
+    "v": 26,
+    "w": 27,
+    "y": 24,
+    "z": 29,
+    "|": 4,
+    "é": 11
+  },
+  "pib": {
+    "'": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "c": 22,
+    "e": 11,
+    "g": 9,
+    "h": 19,
+    "i": 6,
+    "j": 24,
+    "k": 8,
+    "l": 16,
+    "m": 21,
+    "n": 7,
+    "o": 14,
+    "p": 17,
+    "r": 15,
+    "s": 20,
+    "t": 12,
+    "u": 10,
+    "w": 13,
+    "x": 23,
+    "y": 18,
+    "|": 5,
+    "—": 26
+  },
+  "pil": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 22,
+    "d": 21,
+    "e": 10,
+    "f": 31,
+    "g": 30,
+    "h": 35,
+    "i": 14,
+    "j": 28,
+    "k": 13,
+    "l": 17,
+    "m": 12,
+    "n": 7,
+    "o": 26,
+    "p": 27,
+    "r": 8,
+    "s": 18,
+    "t": 15,
+    "u": 29,
+    "v": 34,
+    "w": 19,
+    "y": 23,
+    "z": 33,
+    "|": 4,
+    "à": 37,
+    "è": 36,
+    "ì": 41,
+    "ò": 40,
+    "ù": 42,
+    "ŋ": 24,
+    "ǝ": 38,
+    "ɔ": 25,
+    "ə": 6,
+    "ɛ": 9,
+    "ɣ": 20,
+    "ʋ": 16,
+    "̀": 39
+  },
+  "pir": {
+    "'": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 17,
+    "d": 22,
+    "e": 7,
+    "f": 36,
+    "g": 24,
+    "h": 10,
+    "i": 6,
+    "j": 20,
+    "k": 44,
+    "l": 31,
+    "m": 19,
+    "n": 14,
+    "o": 9,
+    "p": 21,
+    "q": 18,
+    "r": 8,
+    "s": 16,
+    "t": 13,
+    "u": 12,
+    "v": 37,
+    "w": 28,
+    "x": 45,
+    "y": 15,
+    "z": 39,
+    "|": 4,
+    "á": 40,
+    "ã": 26,
+    "é": 38,
+    "í": 35,
+    "ñ": 25,
+    "ó": 43,
+    "õ": 32,
+    "ú": 41,
+    "ĩ": 33,
+    "ũ": 30,
+    "ʉ": 11,
+    "́": 46,
+    "̃": 23,
+    "ẽ": 29,
+    "—": 34
+  },
+  "pis": {
+    "'": 29,
+    "-": 27,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "3": 36,
+    "4": 32,
+    "5": 34,
+    "6": 35,
+    "7": 37,
+    "8": 38,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "d": 20,
+    "e": 6,
+    "f": 16,
+    "g": 14,
+    "h": 17,
+    "i": 9,
+    "j": 26,
+    "k": 15,
+    "l": 11,
+    "m": 8,
+    "n": 10,
+    "o": 7,
+    "p": 22,
+    "r": 21,
+    "s": 13,
+    "t": 12,
+    "u": 18,
+    "v": 25,
+    "w": 24,
+    "y": 23,
+    "|": 4
+  },
+  "pjt": {
+    "'": 37,
+    "-": 27,
+    "0": 30,
+    "1": 32,
+    "2": 35,
+    "3": 41,
+    "4": 36,
+    "5": 39,
+    "6": 40,
+    "7": 34,
+    "8": 43,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 29,
+    "d": 25,
+    "e": 23,
+    "f": 46,
+    "g": 12,
+    "h": 28,
+    "i": 9,
+    "j": 14,
+    "k": 8,
+    "l": 15,
+    "m": 17,
+    "n": 7,
+    "o": 24,
+    "p": 13,
+    "r": 16,
+    "s": 26,
+    "t": 10,
+    "u": 6,
+    "v": 33,
+    "w": 18,
+    "x": 44,
+    "y": 11,
+    "z": 38,
+    "|": 5,
+    "ḻ": 22,
+    "ṉ": 21,
+    "ṟ": 19,
+    "ṯ": 20,
+    "–": 45,
+    "—": 47
+  },
+  "pkb": {
+    "'": 37,
+    "-": 32,
+    "0": 33,
+    "1": 34,
+    "2": 36,
+    "4": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 30,
+    "d": 17,
+    "e": 10,
+    "f": 25,
+    "g": 20,
+    "h": 15,
+    "i": 7,
+    "j": 26,
+    "k": 11,
+    "l": 27,
+    "m": 12,
+    "n": 8,
+    "o": 14,
+    "p": 28,
+    "r": 23,
+    "s": 16,
+    "t": 19,
+    "u": 6,
+    "v": 24,
+    "w": 9,
+    "y": 13,
+    "z": 18,
+    "|": 5,
+    "̯": 22,
+    "ḅ": 31,
+    "ḍ": 29
+  },
+  "pls": {
+    "'": 44,
+    "-": 45,
+    "0": 39,
+    "1": 40,
+    "2": 41,
+    "3": 46,
+    "4": 42,
+    "6": 43,
+    "7": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 23,
+    "c": 17,
+    "d": 19,
+    "e": 8,
+    "f": 34,
+    "g": 21,
+    "h": 9,
+    "i": 7,
+    "j": 11,
+    "k": 16,
+    "l": 27,
+    "m": 25,
+    "n": 5,
+    "o": 10,
+    "p": 30,
+    "q": 38,
+    "r": 15,
+    "s": 20,
+    "t": 13,
+    "u": 12,
+    "v": 37,
+    "x": 18,
+    "y": 26,
+    "z": 36,
+    "|": 4,
+    "á": 24,
+    "é": 28,
+    "í": 32,
+    "ñ": 31,
+    "ó": 35,
+    "ú": 29,
+    "̱": 22,
+    "—": 33,
+    "ꞌ": 14
+  },
+  "plw": {
+    "'": 19,
+    "-": 26,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 33,
+    "4": 31,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "d": 16,
+    "e": 5,
+    "g": 8,
+    "h": 27,
+    "i": 9,
+    "j": 24,
+    "k": 14,
+    "l": 21,
+    "m": 12,
+    "n": 7,
+    "o": 23,
+    "p": 18,
+    "r": 20,
+    "s": 13,
+    "t": 11,
+    "u": 10,
+    "w": 22,
+    "y": 15,
+    "|": 4,
+    "ʼ": 25
+  },
+  "pmf": {
+    "'": 21,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 28,
+    "d": 23,
+    "e": 10,
+    "f": 29,
+    "g": 17,
+    "h": 25,
+    "i": 7,
+    "j": 24,
+    "k": 13,
+    "l": 18,
+    "m": 9,
+    "n": 8,
+    "o": 6,
+    "p": 15,
+    "r": 16,
+    "s": 14,
+    "t": 11,
+    "u": 12,
+    "w": 20,
+    "y": 19,
+    "z": 27,
+    "|": 5
+  },
+  "pny": {
+    "'": 51,
+    "-": 55,
+    "1": 56,
+    "2": 57,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 20,
+    "c": 38,
+    "d": 33,
+    "e": 30,
+    "f": 36,
+    "g": 24,
+    "h": 16,
+    "i": 27,
+    "j": 42,
+    "k": 25,
+    "l": 15,
+    "m": 10,
+    "n": 7,
+    "o": 17,
+    "p": 11,
+    "q": 54,
+    "r": 28,
+    "s": 21,
+    "t": 13,
+    "u": 19,
+    "v": 50,
+    "w": 14,
+    "x": 53,
+    "y": 26,
+    "z": 29,
+    "|": 4,
+    "à": 12,
+    "â": 39,
+    "è": 35,
+    "ì": 37,
+    "î": 48,
+    "ò": 31,
+    "ô": 40,
+    "ù": 22,
+    "û": 43,
+    "ě": 44,
+    "ŋ": 18,
+    "ǎ": 41,
+    "ǐ": 49,
+    "ǒ": 52,
+    "ǔ": 47,
+    "ǹ": 46,
+    "ə": 5,
+    "ɨ": 8,
+    "ʼ": 23,
+    "̀": 6,
+    "̂": 34,
+    "̌": 45,
+    "ẅ": 32
+  },
+  "poh-dialect_eastern": {
+    "'": 7,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 23,
+    "c": 17,
+    "d": 27,
+    "e": 9,
+    "f": 37,
+    "g": 35,
+    "h": 8,
+    "i": 5,
+    "j": 16,
+    "k": 10,
+    "l": 19,
+    "m": 21,
+    "n": 12,
+    "o": 13,
+    "p": 25,
+    "q": 15,
+    "r": 11,
+    "s": 24,
+    "t": 18,
+    "u": 14,
+    "v": 38,
+    "w": 20,
+    "x": 22,
+    "y": 26,
+    "z": 28,
+    "|": 4,
+    "á": 30,
+    "é": 34,
+    "í": 32,
+    "ñ": 39,
+    "ó": 36,
+    "ú": 31,
+    "—": 33
+  },
+  "poh-dialect_western": {
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 8,
+    "d": 28,
+    "e": 10,
+    "h": 9,
+    "i": 5,
+    "j": 14,
+    "k": 15,
+    "l": 17,
+    "m": 21,
+    "n": 12,
+    "o": 13,
+    "p": 26,
+    "q": 25,
+    "r": 11,
+    "s": 24,
+    "t": 18,
+    "u": 16,
+    "w": 19,
+    "x": 23,
+    "y": 27,
+    "z": 30,
+    "|": 4,
+    "̱": 20,
+    "—": 31,
+    "ꞌ": 7
+  },
+  "poi": {
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 32,
+    "c": 11,
+    "d": 22,
+    "e": 18,
+    "f": 38,
+    "g": 25,
+    "h": 30,
+    "i": 6,
+    "j": 10,
+    "k": 40,
+    "l": 31,
+    "m": 7,
+    "n": 20,
+    "o": 13,
+    "p": 12,
+    "q": 33,
+    "r": 28,
+    "s": 17,
+    "t": 9,
+    "u": 15,
+    "v": 37,
+    "w": 24,
+    "x": 29,
+    "y": 14,
+    "z": 39,
+    "|": 4,
+    "á": 27,
+    "é": 21,
+    "í": 34,
+    "ñ": 26,
+    "ó": 36,
+    "ú": 35,
+    "ŋ": 23,
+    "ɨ": 8,
+    "́": 19,
+    "̱": 16
+  },
+  "pol": {
+    "!": 68,
+    "%": 58,
+    "&": 72,
+    "'": 54,
+    "+": 66,
+    ",": 56,
+    "-": 46,
+    ".": 48,
+    "/": 52,
+    "0": 35,
+    "1": 37,
+    "2": 38,
+    "3": 43,
+    "4": 44,
+    "5": 41,
+    "6": 45,
+    "7": 50,
+    "8": 47,
+    "9": 40,
+    ":": 51,
+    ";": 70,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 15,
+    "d": 17,
+    "e": 8,
+    "f": 34,
+    "g": 24,
+    "h": 28,
+    "i": 6,
+    "j": 22,
+    "k": 18,
+    "l": 23,
+    "m": 20,
+    "n": 9,
+    "o": 7,
+    "p": 19,
+    "q": 55,
+    "r": 11,
+    "s": 13,
+    "t": 14,
+    "u": 21,
+    "v": 42,
+    "w": 12,
+    "x": 49,
+    "y": 16,
+    "z": 10,
+    "|": 4,
+    "°": 67,
+    "²": 62,
+    "á": 65,
+    "ã": 64,
+    "ç": 69,
+    "é": 63,
+    "í": 71,
+    "ó": 30,
+    "õ": 60,
+    "ú": 61,
+    "ü": 59,
+    "ą": 27,
+    "ć": 33,
+    "ę": 29,
+    "ł": 25,
+    "ń": 36,
+    "ś": 32,
+    "ź": 39,
+    "ż": 31,
+    "–": 57,
+    "—": 73,
+    "”": 53
+  },
+  "por": {
+    "!": 76,
+    "\"": 60,
+    "$": 64,
+    "%": 59,
+    "&": 67,
+    "'": 58,
+    "+": 68,
+    ",": 56,
+    "-": 38,
+    ".": 48,
+    "/": 55,
+    "0": 35,
+    "1": 37,
+    "2": 41,
+    "3": 50,
+    "4": 49,
+    "5": 47,
+    "6": 51,
+    "7": 54,
+    "8": 52,
+    "9": 46,
+    ":": 57,
+    ";": 63,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 74,
+    "]": 75,
+    "a": 5,
+    "b": 22,
+    "c": 16,
+    "d": 12,
+    "e": 6,
+    "f": 21,
+    "g": 20,
+    "h": 23,
+    "i": 10,
+    "j": 31,
+    "k": 39,
+    "l": 18,
+    "m": 14,
+    "n": 11,
+    "o": 7,
+    "p": 17,
+    "q": 24,
+    "r": 9,
+    "s": 8,
+    "t": 13,
+    "u": 15,
+    "v": 19,
+    "w": 45,
+    "x": 32,
+    "y": 44,
+    "z": 30,
+    "|": 4,
+    "ª": 69,
+    "°": 62,
+    "²": 70,
+    "º": 61,
+    "à": 43,
+    "á": 28,
+    "â": 42,
+    "ã": 25,
+    "ç": 26,
+    "é": 27,
+    "ê": 33,
+    "ë": 71,
+    "í": 29,
+    "ò": 78,
+    "ó": 34,
+    "ô": 53,
+    "õ": 36,
+    "ú": 40,
+    "ü": 77,
+    "ž": 80,
+    "́": 79,
+    "​": 66,
+    "—": 81,
+    "‘": 72,
+    "’": 73,
+    "”": 65
+  },
+  "poy": {
+    "'": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 29,
+    "d": 22,
+    "e": 13,
+    "f": 26,
+    "g": 15,
+    "h": 20,
+    "i": 6,
+    "j": 27,
+    "k": 11,
+    "l": 12,
+    "m": 10,
+    "n": 8,
+    "o": 19,
+    "p": 21,
+    "r": 16,
+    "s": 18,
+    "t": 14,
+    "u": 7,
+    "v": 25,
+    "w": 9,
+    "y": 17,
+    "z": 24,
+    "|": 5
+  },
+  "ppk": {
+    "'": 8,
+    "-": 16,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 27,
+    "d": 24,
+    "e": 12,
+    "f": 29,
+    "g": 20,
+    "h": 18,
+    "i": 7,
+    "j": 26,
+    "k": 15,
+    "l": 17,
+    "m": 13,
+    "n": 11,
+    "o": 6,
+    "p": 14,
+    "r": 19,
+    "s": 21,
+    "t": 9,
+    "u": 10,
+    "w": 23,
+    "y": 25,
+    "z": 30,
+    "|": 5,
+    "–": 28
+  },
+  "pps": {
+    "'": 38,
+    "-": 39,
+    "1": 45,
+    "2": 42,
+    "5": 47,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 41,
+    "a": 6,
+    "b": 28,
+    "c": 17,
+    "d": 18,
+    "e": 8,
+    "f": 34,
+    "g": 21,
+    "h": 19,
+    "i": 7,
+    "j": 11,
+    "k": 13,
+    "l": 32,
+    "m": 23,
+    "n": 5,
+    "o": 9,
+    "p": 33,
+    "q": 37,
+    "r": 20,
+    "s": 15,
+    "t": 10,
+    "u": 12,
+    "v": 36,
+    "x": 16,
+    "y": 29,
+    "z": 35,
+    "|": 4,
+    "à": 46,
+    "á": 26,
+    "é": 24,
+    "ì": 44,
+    "í": 25,
+    "ñ": 30,
+    "ó": 31,
+    "ù": 43,
+    "ú": 27,
+    "̱": 22,
+    "—": 48,
+    "ꞌ": 14
+  },
+  "prf": {
+    "'": 28,
+    "-": 26,
+    "0": 33,
+    "1": 34,
+    "2": 35,
+    "4": 37,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 27,
+    "d": 12,
+    "e": 10,
+    "f": 29,
+    "g": 15,
+    "h": 22,
+    "i": 7,
+    "j": 25,
+    "k": 9,
+    "l": 18,
+    "m": 14,
+    "n": 6,
+    "o": 19,
+    "p": 17,
+    "q": 32,
+    "r": 23,
+    "s": 16,
+    "t": 8,
+    "u": 11,
+    "v": 31,
+    "w": 24,
+    "x": 36,
+    "y": 13,
+    "z": 30,
+    "|": 4,
+    "á": 20
+  },
+  "prk": {
+    "-": 30,
+    "2": 33,
+    "4": 32,
+    "6": 34,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 23,
+    "d": 24,
+    "e": 10,
+    "f": 28,
+    "g": 14,
+    "h": 7,
+    "i": 6,
+    "j": 27,
+    "k": 9,
+    "l": 22,
+    "m": 13,
+    "n": 8,
+    "o": 17,
+    "p": 16,
+    "r": 20,
+    "s": 19,
+    "t": 18,
+    "u": 12,
+    "v": 26,
+    "w": 15,
+    "x": 11,
+    "y": 21,
+    "z": 29,
+    "|": 4
+  },
+  "prt": {
+    "'": 57,
+    "-": 51,
+    "0": 49,
+    "1": 44,
+    "2": 45,
+    "3": 46,
+    "4": 47,
+    "5": 50,
+    "6": 52,
+    "7": 54,
+    "8": 56,
+    "9": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ก": 10,
+    "ค": 24,
+    "ง": 11,
+    "จ": 26,
+    "ซ": 17,
+    "ญ": 31,
+    "ด": 22,
+    "ต": 21,
+    "ท": 38,
+    "น": 7,
+    "บ": 32,
+    "ป": 27,
+    "พ": 41,
+    "ฟ": 43,
+    "ม": 15,
+    "ย": 28,
+    "ร": 23,
+    "ล": 12,
+    "ว": 25,
+    "ห": 29,
+    "อ": 5,
+    "ฮ": 39,
+    "ะ": 18,
+    "ั": 9,
+    "า": 8,
+    "ิ": 20,
+    "ี": 19,
+    "ึ": 40,
+    "ื": 35,
+    "ุ": 33,
+    "ู": 36,
+    "เ": 6,
+    "แ": 16,
+    "โ": 37,
+    "ใ": 42,
+    "ไ": 34,
+    "็": 13,
+    "้": 53,
+    "๊": 48,
+    "ํ": 30,
+    "‍": 14
+  },
+  "pse": {
+    "'": 24,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 23,
+    "d": 12,
+    "e": 11,
+    "g": 10,
+    "h": 21,
+    "i": 8,
+    "j": 20,
+    "k": 9,
+    "l": 15,
+    "m": 14,
+    "n": 7,
+    "p": 19,
+    "r": 18,
+    "s": 16,
+    "t": 13,
+    "u": 6,
+    "w": 26,
+    "y": 22,
+    "|": 5
+  },
+  "pss": {
+    "'": 25,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 24,
+    "d": 23,
+    "e": 11,
+    "g": 15,
+    "h": 19,
+    "i": 6,
+    "k": 13,
+    "l": 18,
+    "m": 8,
+    "n": 5,
+    "o": 10,
+    "p": 14,
+    "r": 21,
+    "s": 16,
+    "t": 12,
+    "u": 9,
+    "w": 17,
+    "y": 20,
+    "|": 4,
+    "–": 26
+  },
+  "ptu": {
+    "'": 14,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 30,
+    "d": 18,
+    "e": 16,
+    "f": 28,
+    "g": 19,
+    "h": 21,
+    "i": 7,
+    "j": 26,
+    "k": 15,
+    "l": 10,
+    "m": 9,
+    "n": 6,
+    "o": 11,
+    "p": 17,
+    "r": 25,
+    "s": 13,
+    "t": 12,
+    "u": 8,
+    "w": 27,
+    "y": 23,
+    "z": 29,
+    "|": 5,
+    "ä": 22
+  },
+  "pui": {
+    "-": 25,
+    "0": 40,
+    "4": 42,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 18,
+    "d": 20,
+    "e": 11,
+    "f": 33,
+    "g": 13,
+    "h": 22,
+    "i": 6,
+    "j": 7,
+    "k": 39,
+    "l": 21,
+    "m": 17,
+    "n": 12,
+    "o": 9,
+    "p": 23,
+    "q": 29,
+    "r": 8,
+    "s": 15,
+    "t": 10,
+    "u": 14,
+    "v": 35,
+    "w": 24,
+    "x": 38,
+    "y": 16,
+    "z": 36,
+    "|": 5,
+    "á": 26,
+    "é": 32,
+    "í": 30,
+    "ñ": 37,
+    "ó": 34,
+    "ú": 27,
+    "ü": 31,
+    "ŕ": 28
+  },
+  "pus": {
+    "!": 113,
+    "\"": 104,
+    "$": 118,
+    "%": 128,
+    "&": 136,
+    "'": 109,
+    "+": 123,
+    ",": 103,
+    "-": 89,
+    ".": 67,
+    "/": 98,
+    "0": 52,
+    "1": 54,
+    "2": 64,
+    "3": 80,
+    "4": 76,
+    "5": 73,
+    "6": 81,
+    "7": 85,
+    "8": 78,
+    "9": 69,
+    ":": 84,
+    ";": 126,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "_": 110,
+    "`": 132,
+    "a": 39,
+    "b": 74,
+    "c": 60,
+    "d": 62,
+    "e": 40,
+    "f": 82,
+    "g": 65,
+    "h": 61,
+    "i": 47,
+    "j": 87,
+    "k": 71,
+    "l": 53,
+    "m": 66,
+    "n": 45,
+    "o": 46,
+    "p": 68,
+    "q": 102,
+    "r": 48,
+    "s": 49,
+    "t": 51,
+    "u": 58,
+    "v": 77,
+    "w": 86,
+    "x": 94,
+    "y": 75,
+    "z": 90,
+    "|": 4,
+    "¥": 135,
+    "°": 124,
+    "é": 127,
+    "í": 133,
+    "،": 43,
+    "؛": 121,
+    "؟": 131,
+    "ء": 125,
+    "آ": 63,
+    "أ": 112,
+    "ؤ": 100,
+    "ئ": 55,
+    "ا": 6,
+    "ب": 20,
+    "ت": 15,
+    "ث": 57,
+    "ج": 31,
+    "ح": 36,
+    "خ": 21,
+    "د": 10,
+    "ذ": 79,
+    "ر": 9,
+    "ز": 26,
+    "س": 19,
+    "ش": 22,
+    "ص": 42,
+    "ض": 56,
+    "ط": 44,
+    "ظ": 59,
+    "ع": 33,
+    "غ": 28,
+    "ـ": 120,
+    "ف": 29,
+    "ق": 35,
+    "ك": 107,
+    "ل": 12,
+    "م": 16,
+    "ن": 11,
+    "ه": 7,
+    "و": 5,
+    "ى": 111,
+    "ي": 17,
+    "ً": 92,
+    "ٌ": 137,
+    "َ": 105,
+    "ُ": 134,
+    "ّ": 116,
+    "٪": 122,
+    "ټ": 27,
+    "پ": 18,
+    "ځ": 30,
+    "څ": 34,
+    "چ": 24,
+    "ډ": 32,
+    "ړ": 23,
+    "ږ": 38,
+    "ژ": 50,
+    "ښ": 37,
+    "ک": 13,
+    "ګ": 25,
+    "گ": 129,
+    "ڼ": 70,
+    "ھ": 117,
+    "ی": 8,
+    "ۍ": 41,
+    "ې": 14,
+    "ے": 130,
+    "۔": 91,
+    "۰": 72,
+    "۱": 83,
+    "۲": 88,
+    "۳": 93,
+    "۴": 96,
+    "۵": 97,
+    "۶": 95,
+    "۷": 106,
+    "۸": 101,
+    "۹": 99,
+    "‌": 115,
+    "‍": 138,
+    "–": 108,
+    "—": 119,
+    "”": 114
+  },
+  "pwg": {
+    "'": 40,
+    "0": 28,
+    "1": 26,
+    "2": 27,
+    "3": 33,
+    "4": 30,
+    "5": 31,
+    "6": 35,
+    "7": 38,
+    "8": 34,
+    "9": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 32,
+    "d": 19,
+    "e": 10,
+    "f": 37,
+    "g": 21,
+    "h": 23,
+    "i": 6,
+    "j": 25,
+    "k": 8,
+    "l": 24,
+    "m": 11,
+    "n": 7,
+    "o": 9,
+    "p": 22,
+    "r": 14,
+    "s": 20,
+    "t": 15,
+    "u": 12,
+    "v": 13,
+    "w": 18,
+    "x": 39,
+    "y": 17,
+    "z": 29,
+    "|": 5
+  },
+  "pww": {
+    "'": 43,
+    "-": 37,
+    "1": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ก": 21,
+    "ค": 30,
+    "ฆ": 35,
+    "ง": 6,
+    "จ": 32,
+    "ช": 31,
+    "ซ": 20,
+    "ฌ": 26,
+    "ด": 27,
+    "ต": 40,
+    "ท": 17,
+    "น": 16,
+    "บ": 25,
+    "ป": 39,
+    "ผ": 48,
+    "พ": 18,
+    "ฟ": 44,
+    "ม": 19,
+    "ย": 28,
+    "ร": 41,
+    "ล": 8,
+    "ว": 23,
+    "ส": 47,
+    "อ": 7,
+    "ฮ": 42,
+    "ะ": 15,
+    "ั": 45,
+    "า": 14,
+    "ิ": 36,
+    "ี": 13,
+    "ึ": 38,
+    "ื": 33,
+    "ุ": 34,
+    "ู": 24,
+    "เ": 9,
+    "แ": 10,
+    "โ": 22,
+    "ไ": 29,
+    "่": 11,
+    "้": 5,
+    "๊": 12
+  },
+  "pxm": {
+    "'": 12,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "c": 29,
+    "d": 27,
+    "e": 21,
+    "f": 31,
+    "g": 30,
+    "h": 32,
+    "i": 13,
+    "j": 9,
+    "k": 11,
+    "l": 24,
+    "m": 14,
+    "n": 10,
+    "o": 17,
+    "p": 15,
+    "q": 38,
+    "r": 23,
+    "s": 16,
+    "t": 6,
+    "u": 18,
+    "v": 35,
+    "w": 22,
+    "x": 20,
+    "y": 8,
+    "z": 37,
+    "|": 4,
+    "á": 36,
+    "ä": 19,
+    "é": 28,
+    "ë": 5,
+    "í": 33,
+    "ñ": 39,
+    "ó": 34,
+    "ú": 25
+  },
+  "qub": {
+    "'": 41,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 7,
+    "d": 25,
+    "e": 23,
+    "f": 32,
+    "g": 19,
+    "h": 10,
+    "i": 8,
+    "j": 18,
+    "k": 43,
+    "l": 17,
+    "m": 14,
+    "n": 6,
+    "o": 20,
+    "p": 16,
+    "q": 24,
+    "r": 13,
+    "s": 12,
+    "t": 15,
+    "u": 9,
+    "v": 38,
+    "w": 21,
+    "x": 42,
+    "y": 11,
+    "z": 37,
+    "|": 5,
+    "á": 40,
+    "ä": 22,
+    "é": 36,
+    "ë": 30,
+    "í": 35,
+    "ï": 29,
+    "ñ": 28,
+    "ó": 39,
+    "ö": 26,
+    "ú": 33,
+    "ü": 34
+  },
+  "quc-dialect_central": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 7,
+    "d": 28,
+    "e": 10,
+    "f": 38,
+    "g": 33,
+    "h": 14,
+    "i": 5,
+    "j": 12,
+    "k": 18,
+    "l": 17,
+    "m": 20,
+    "n": 13,
+    "o": 16,
+    "p": 26,
+    "q": 24,
+    "r": 8,
+    "s": 23,
+    "t": 15,
+    "u": 9,
+    "v": 37,
+    "w": 21,
+    "x": 19,
+    "y": 29,
+    "z": 27,
+    "|": 4,
+    "á": 35,
+    "ä": 25,
+    "é": 34,
+    "í": 32,
+    "ñ": 39,
+    "ó": 36,
+    "ú": 31,
+    "—": 30,
+    "ꞌ": 11
+  },
+  "quc-dialect_east": {
+    "'": 10,
+    "-": 33,
+    "0": 36,
+    "1": 35,
+    "2": 37,
+    "3": 43,
+    "4": 39,
+    "5": 42,
+    "6": 38,
+    "7": 40,
+    "8": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 14,
+    "d": 28,
+    "e": 9,
+    "f": 30,
+    "g": 29,
+    "h": 13,
+    "i": 6,
+    "j": 12,
+    "k": 8,
+    "l": 17,
+    "m": 23,
+    "n": 15,
+    "o": 18,
+    "p": 25,
+    "q": 19,
+    "r": 7,
+    "s": 24,
+    "t": 16,
+    "u": 11,
+    "v": 32,
+    "w": 20,
+    "x": 22,
+    "y": 26,
+    "z": 27,
+    "|": 4,
+    "ä": 34,
+    "ñ": 41,
+    "–": 31,
+    "—": 45
+  },
+  "quc-dialect_north": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 11,
+    "d": 28,
+    "e": 10,
+    "f": 34,
+    "g": 31,
+    "h": 12,
+    "i": 7,
+    "j": 9,
+    "k": 8,
+    "l": 13,
+    "m": 21,
+    "n": 15,
+    "o": 18,
+    "p": 25,
+    "q": 17,
+    "r": 16,
+    "s": 24,
+    "t": 19,
+    "u": 14,
+    "v": 35,
+    "w": 23,
+    "x": 20,
+    "y": 26,
+    "z": 27,
+    "|": 4,
+    "á": 36,
+    "é": 30,
+    "í": 33,
+    "ñ": 37,
+    "ó": 32,
+    "ú": 29
+  },
+  "quf": {
+    "'": 22,
+    "-": 27,
+    "0": 33,
+    "1": 35,
+    "2": 36,
+    "4": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 19,
+    "d": 21,
+    "e": 23,
+    "f": 30,
+    "g": 26,
+    "h": 12,
+    "i": 7,
+    "j": 24,
+    "k": 9,
+    "l": 11,
+    "m": 14,
+    "n": 6,
+    "o": 28,
+    "p": 13,
+    "q": 16,
+    "r": 15,
+    "s": 17,
+    "t": 18,
+    "u": 8,
+    "v": 31,
+    "w": 20,
+    "x": 34,
+    "y": 10,
+    "z": 32,
+    "|": 5,
+    "ñ": 29
+  },
+  "quh": {
+    "'": 24,
+    "0": 36,
+    "1": 39,
+    "2": 40,
+    "3": 43,
+    "4": 41,
+    "5": 42,
+    "6": 44,
+    "7": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 29,
+    "c": 15,
+    "d": 25,
+    "e": 20,
+    "f": 34,
+    "g": 30,
+    "h": 12,
+    "i": 7,
+    "j": 13,
+    "k": 9,
+    "l": 22,
+    "m": 18,
+    "n": 6,
+    "o": 21,
+    "p": 16,
+    "q": 17,
+    "r": 19,
+    "s": 10,
+    "t": 11,
+    "u": 8,
+    "v": 31,
+    "w": 23,
+    "x": 38,
+    "y": 14,
+    "z": 33,
+    "|": 5,
+    "á": 28,
+    "é": 35,
+    "í": 27,
+    "ñ": 26,
+    "ó": 37,
+    "ú": 32
+  },
+  "qul": {
+    "'": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 12,
+    "d": 25,
+    "e": 24,
+    "f": 30,
+    "g": 29,
+    "h": 11,
+    "i": 7,
+    "j": 21,
+    "k": 10,
+    "l": 19,
+    "m": 13,
+    "n": 6,
+    "o": 23,
+    "p": 16,
+    "q": 9,
+    "r": 18,
+    "s": 15,
+    "t": 17,
+    "u": 8,
+    "v": 33,
+    "w": 20,
+    "x": 40,
+    "y": 14,
+    "z": 36,
+    "|": 5,
+    "á": 38,
+    "ä": 35,
+    "é": 32,
+    "í": 31,
+    "ï": 39,
+    "ñ": 26,
+    "ó": 34,
+    "ú": 37,
+    "–": 27
+  },
+  "quw": {
+    "'": 42,
+    "-": 44,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 9,
+    "d": 19,
+    "e": 23,
+    "f": 29,
+    "g": 17,
+    "h": 10,
+    "i": 6,
+    "j": 18,
+    "k": 43,
+    "l": 15,
+    "m": 13,
+    "n": 8,
+    "o": 22,
+    "p": 14,
+    "q": 24,
+    "r": 12,
+    "s": 11,
+    "t": 16,
+    "u": 7,
+    "v": 27,
+    "x": 39,
+    "y": 20,
+    "z": 26,
+    "|": 5,
+    "á": 34,
+    "ä": 45,
+    "é": 32,
+    "í": 30,
+    "ï": 40,
+    "ñ": 21,
+    "ó": 35,
+    "ú": 28,
+    "ü": 33,
+    "ź": 31,
+    "ḿ": 46,
+    "—": 41
+  },
+  "quy": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 15,
+    "d": 24,
+    "e": 21,
+    "f": 31,
+    "g": 29,
+    "h": 14,
+    "i": 7,
+    "j": 25,
+    "k": 8,
+    "l": 19,
+    "m": 13,
+    "n": 6,
+    "o": 20,
+    "p": 11,
+    "q": 10,
+    "r": 18,
+    "s": 12,
+    "t": 17,
+    "u": 9,
+    "v": 30,
+    "w": 22,
+    "x": 35,
+    "y": 16,
+    "z": 33,
+    "|": 5,
+    "á": 28,
+    "é": 36,
+    "í": 34,
+    "ñ": 23,
+    "ó": 37,
+    "ú": 38,
+    "—": 27
+  },
+  "quz": {
+    "'": 24,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 15,
+    "d": 25,
+    "e": 22,
+    "f": 32,
+    "g": 27,
+    "h": 12,
+    "i": 7,
+    "j": 26,
+    "k": 10,
+    "l": 19,
+    "m": 17,
+    "n": 6,
+    "o": 20,
+    "p": 13,
+    "q": 9,
+    "r": 18,
+    "s": 11,
+    "t": 16,
+    "u": 8,
+    "v": 31,
+    "w": 21,
+    "x": 39,
+    "y": 14,
+    "z": 34,
+    "|": 5,
+    "á": 36,
+    "é": 35,
+    "í": 33,
+    "ñ": 23,
+    "ó": 37,
+    "ú": 38,
+    "–": 30
+  },
+  "qvc": {
+    "'": 39,
+    "0": 30,
+    "1": 37,
+    "2": 33,
+    "3": 40,
+    "4": 32,
+    "5": 38,
+    "6": 44,
+    "7": 43,
+    "8": 41,
+    "9": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 18,
+    "d": 20,
+    "e": 29,
+    "f": 28,
+    "g": 23,
+    "h": 9,
+    "i": 6,
+    "j": 24,
+    "k": 13,
+    "l": 16,
+    "m": 15,
+    "n": 7,
+    "o": 25,
+    "p": 14,
+    "q": 12,
+    "r": 17,
+    "s": 11,
+    "t": 19,
+    "u": 8,
+    "w": 21,
+    "y": 10,
+    "|": 5,
+    "á": 31,
+    "í": 36,
+    "ñ": 26,
+    "˻": 34,
+    "˼": 35,
+    "—": 27
+  },
+  "qve": {
+    "'": 27,
+    "-": 37,
+    "0": 40,
+    "1": 41,
+    "2": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 14,
+    "d": 23,
+    "e": 21,
+    "f": 30,
+    "g": 28,
+    "h": 12,
+    "i": 7,
+    "j": 25,
+    "k": 10,
+    "l": 19,
+    "m": 18,
+    "n": 6,
+    "o": 20,
+    "p": 13,
+    "q": 11,
+    "r": 17,
+    "s": 9,
+    "t": 15,
+    "u": 8,
+    "v": 29,
+    "w": 22,
+    "x": 39,
+    "y": 16,
+    "z": 33,
+    "|": 5,
+    "á": 31,
+    "é": 36,
+    "í": 34,
+    "ñ": 24,
+    "ó": 35,
+    "ú": 38,
+    "—": 32
+  },
+  "qvh": {
+    "'": 42,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 10,
+    "d": 25,
+    "e": 21,
+    "f": 34,
+    "g": 16,
+    "h": 17,
+    "i": 7,
+    "j": 24,
+    "k": 43,
+    "l": 20,
+    "m": 15,
+    "n": 6,
+    "o": 18,
+    "p": 13,
+    "q": 23,
+    "r": 14,
+    "s": 9,
+    "t": 12,
+    "u": 8,
+    "v": 29,
+    "w": 22,
+    "x": 41,
+    "y": 11,
+    "z": 35,
+    "|": 5,
+    "á": 33,
+    "ä": 19,
+    "é": 37,
+    "ë": 28,
+    "í": 39,
+    "ï": 31,
+    "ñ": 38,
+    "ó": 40,
+    "ö": 26,
+    "ú": 32,
+    "ü": 36
+  },
+  "qvm": {
+    "'": 42,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 8,
+    "d": 25,
+    "e": 20,
+    "f": 29,
+    "g": 12,
+    "h": 17,
+    "i": 7,
+    "j": 24,
+    "k": 43,
+    "l": 19,
+    "m": 16,
+    "n": 6,
+    "o": 18,
+    "p": 14,
+    "q": 23,
+    "r": 15,
+    "s": 10,
+    "t": 11,
+    "u": 9,
+    "v": 30,
+    "w": 22,
+    "x": 41,
+    "y": 13,
+    "z": 34,
+    "|": 5,
+    "á": 35,
+    "ä": 21,
+    "é": 37,
+    "ë": 28,
+    "í": 39,
+    "ï": 32,
+    "ñ": 38,
+    "ó": 40,
+    "ö": 27,
+    "ú": 33,
+    "ü": 36
+  },
+  "qvn": {
+    "'": 39,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 26,
+    "c": 7,
+    "d": 24,
+    "e": 23,
+    "f": 33,
+    "g": 13,
+    "h": 14,
+    "i": 9,
+    "j": 22,
+    "k": 44,
+    "l": 19,
+    "m": 12,
+    "n": 6,
+    "o": 20,
+    "p": 16,
+    "q": 25,
+    "r": 10,
+    "s": 15,
+    "t": 17,
+    "u": 8,
+    "v": 30,
+    "w": 21,
+    "x": 42,
+    "y": 11,
+    "z": 31,
+    "|": 5,
+    "á": 40,
+    "ä": 18,
+    "é": 36,
+    "ë": 32,
+    "í": 38,
+    "ï": 28,
+    "ñ": 34,
+    "ó": 41,
+    "ö": 35,
+    "ú": 37,
+    "ü": 27,
+    "—": 43
+  },
+  "qvo": {
+    "'": 31,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 5,
+    "d": 23,
+    "e": 22,
+    "f": 28,
+    "g": 27,
+    "h": 10,
+    "i": 7,
+    "j": 21,
+    "k": 36,
+    "l": 16,
+    "m": 15,
+    "n": 9,
+    "o": 19,
+    "p": 12,
+    "q": 18,
+    "r": 14,
+    "s": 11,
+    "t": 13,
+    "u": 8,
+    "v": 29,
+    "x": 33,
+    "y": 17,
+    "z": 30,
+    "|": 6,
+    "á": 35,
+    "é": 38,
+    "í": 34,
+    "ñ": 20,
+    "ó": 32,
+    "ú": 37,
+    "—": 24
+  },
+  "qvs": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 15,
+    "d": 21,
+    "e": 20,
+    "f": 29,
+    "g": 26,
+    "h": 10,
+    "i": 8,
+    "j": 23,
+    "k": 7,
+    "l": 17,
+    "m": 18,
+    "n": 6,
+    "o": 22,
+    "p": 13,
+    "r": 16,
+    "s": 11,
+    "t": 12,
+    "u": 9,
+    "v": 27,
+    "w": 19,
+    "x": 32,
+    "y": 14,
+    "z": 33,
+    "|": 5,
+    "á": 31,
+    "í": 30,
+    "ñ": 24,
+    "—": 28
+  },
+  "qvw": {
+    "'": 28,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 33,
+    "c": 6,
+    "d": 26,
+    "e": 31,
+    "f": 35,
+    "g": 34,
+    "h": 12,
+    "i": 8,
+    "j": 24,
+    "k": 45,
+    "l": 10,
+    "m": 13,
+    "n": 7,
+    "o": 29,
+    "p": 14,
+    "q": 20,
+    "r": 22,
+    "s": 16,
+    "t": 15,
+    "u": 9,
+    "v": 38,
+    "w": 18,
+    "x": 44,
+    "y": 11,
+    "z": 39,
+    "|": 5,
+    "á": 30,
+    "ä": 17,
+    "é": 41,
+    "ë": 40,
+    "í": 37,
+    "ï": 27,
+    "ñ": 25,
+    "ó": 46,
+    "ö": 48,
+    "ú": 36,
+    "ü": 23,
+    "ć": 19,
+    "ś": 21,
+    "˻": 42,
+    "˼": 43,
+    "—": 47
+  },
+  "qvz": {
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 8,
+    "d": 19,
+    "e": 23,
+    "f": 29,
+    "g": 15,
+    "h": 10,
+    "i": 6,
+    "j": 25,
+    "k": 37,
+    "l": 16,
+    "m": 14,
+    "n": 9,
+    "o": 20,
+    "p": 17,
+    "q": 24,
+    "r": 13,
+    "s": 11,
+    "t": 12,
+    "u": 7,
+    "v": 28,
+    "x": 36,
+    "y": 18,
+    "z": 26,
+    "|": 5,
+    "á": 27,
+    "é": 35,
+    "í": 32,
+    "ñ": 21,
+    "ó": 33,
+    "ú": 34,
+    "ü": 31
+  },
+  "qwh": {
+    "'": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 9,
+    "d": 23,
+    "e": 14,
+    "f": 30,
+    "g": 27,
+    "h": 20,
+    "i": 7,
+    "j": 24,
+    "l": 17,
+    "m": 15,
+    "n": 6,
+    "o": 18,
+    "p": 16,
+    "q": 10,
+    "r": 13,
+    "s": 11,
+    "t": 12,
+    "u": 8,
+    "v": 35,
+    "w": 21,
+    "x": 38,
+    "y": 19,
+    "z": 34,
+    "|": 5,
+    "ä": 22,
+    "ë": 26,
+    "ï": 28,
+    "ñ": 31,
+    "ö": 32,
+    "ü": 33,
+    "˻": 36,
+    "˼": 37,
+    "—": 29
+  },
+  "qxh": {
+    "-": 27,
+    "2": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 33,
+    "c": 9,
+    "d": 24,
+    "e": 26,
+    "f": 36,
+    "g": 22,
+    "h": 10,
+    "i": 7,
+    "j": 18,
+    "k": 42,
+    "l": 17,
+    "m": 15,
+    "n": 6,
+    "o": 25,
+    "p": 13,
+    "q": 20,
+    "r": 12,
+    "s": 11,
+    "t": 14,
+    "u": 8,
+    "v": 40,
+    "w": 19,
+    "x": 41,
+    "y": 16,
+    "z": 39,
+    "|": 5,
+    "­": 44,
+    "á": 38,
+    "ä": 21,
+    "ç": 30,
+    "é": 34,
+    "í": 35,
+    "ï": 32,
+    "ñ": 28,
+    "ó": 37,
+    "ú": 29,
+    "ü": 31,
+    "ć": 23
+  },
+  "qxl": {
+    "'": 25,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 16,
+    "d": 13,
+    "e": 24,
+    "f": 31,
+    "g": 12,
+    "h": 9,
+    "i": 6,
+    "j": 18,
+    "k": 11,
+    "l": 15,
+    "m": 14,
+    "n": 8,
+    "o": 23,
+    "p": 20,
+    "q": 36,
+    "r": 17,
+    "s": 10,
+    "t": 19,
+    "u": 7,
+    "v": 29,
+    "w": 27,
+    "x": 38,
+    "y": 22,
+    "z": 28,
+    "|": 5,
+    "á": 34,
+    "é": 32,
+    "í": 33,
+    "ñ": 26,
+    "ó": 35,
+    "ú": 30,
+    "ü": 39,
+    "–": 40
+  },
+  "qxn": {
+    "'": 42,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 9,
+    "d": 25,
+    "e": 19,
+    "f": 30,
+    "g": 14,
+    "h": 20,
+    "i": 7,
+    "j": 26,
+    "k": 41,
+    "l": 17,
+    "m": 15,
+    "n": 6,
+    "o": 18,
+    "p": 16,
+    "q": 23,
+    "r": 13,
+    "s": 10,
+    "t": 12,
+    "u": 8,
+    "v": 31,
+    "w": 22,
+    "x": 40,
+    "y": 11,
+    "z": 34,
+    "|": 5,
+    "á": 38,
+    "ä": 21,
+    "é": 35,
+    "ë": 32,
+    "í": 36,
+    "ï": 29,
+    "ñ": 27,
+    "ó": 37,
+    "ö": 24,
+    "ú": 39,
+    "ü": 33
+  },
+  "qxo": {
+    "'": 37,
+    "-": 35,
+    "0": 36,
+    "1": 39,
+    "2": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 9,
+    "d": 25,
+    "e": 21,
+    "f": 32,
+    "g": 15,
+    "h": 19,
+    "i": 7,
+    "j": 26,
+    "k": 41,
+    "l": 17,
+    "m": 14,
+    "n": 6,
+    "o": 20,
+    "p": 16,
+    "q": 24,
+    "r": 13,
+    "s": 12,
+    "t": 10,
+    "u": 8,
+    "v": 30,
+    "w": 23,
+    "x": 38,
+    "y": 11,
+    "z": 34,
+    "|": 5,
+    "ä": 18,
+    "ë": 31,
+    "ï": 29,
+    "ñ": 27,
+    "ö": 22,
+    "ü": 33
+  },
+  "qxr": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "c": 5,
+    "d": 20,
+    "e": 19,
+    "f": 30,
+    "g": 21,
+    "h": 9,
+    "i": 7,
+    "j": 22,
+    "k": 38,
+    "l": 15,
+    "m": 13,
+    "n": 10,
+    "o": 17,
+    "p": 12,
+    "q": 24,
+    "r": 16,
+    "s": 11,
+    "t": 14,
+    "u": 8,
+    "v": 28,
+    "x": 37,
+    "y": 18,
+    "z": 26,
+    "|": 6,
+    "á": 34,
+    "é": 33,
+    "í": 31,
+    "ñ": 23,
+    "ó": 35,
+    "ú": 29,
+    "ü": 36,
+    "—": 27
+  },
+  "rah": {
+    "'": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ং": 13,
+    "ঃ": 41,
+    "অ": 31,
+    "আ": 21,
+    "ই": 33,
+    "উ": 24,
+    "এ": 38,
+    "ও": 39,
+    "ক": 8,
+    "খ": 28,
+    "গ": 30,
+    "ঙ": 17,
+    "চ": 25,
+    "ছ": 16,
+    "জ": 23,
+    "ত": 11,
+    "থ": 35,
+    "দ": 26,
+    "ন": 6,
+    "প": 27,
+    "ফ": 34,
+    "ব": 12,
+    "ভ": 42,
+    "ম": 18,
+    "য": 14,
+    "ল": 32,
+    "হ": 36,
+    "়": 15,
+    "া": 4,
+    "ি": 10,
+    "ু": 22,
+    "ে": 9,
+    "ো": 19,
+    "্": 29,
+    "ৗ": 20,
+    "ৰ": 7,
+    "ৱ": 37
+  },
+  "rai": {
+    "'": 30,
+    "-": 33,
+    "0": 24,
+    "1": 22,
+    "2": 23,
+    "3": 29,
+    "4": 26,
+    "5": 28,
+    "6": 27,
+    "7": 25,
+    "8": 31,
+    "9": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "d": 15,
+    "e": 20,
+    "f": 34,
+    "g": 16,
+    "i": 6,
+    "k": 12,
+    "l": 18,
+    "m": 10,
+    "n": 7,
+    "o": 14,
+    "p": 13,
+    "r": 11,
+    "s": 21,
+    "t": 9,
+    "u": 8,
+    "w": 19,
+    "|": 5
+  },
+  "rap": {
+    "'": 30,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 37,
+    "c": 29,
+    "d": 34,
+    "e": 6,
+    "f": 26,
+    "g": 39,
+    "h": 13,
+    "i": 8,
+    "j": 33,
+    "k": 14,
+    "l": 31,
+    "m": 15,
+    "n": 16,
+    "o": 10,
+    "p": 19,
+    "r": 12,
+    "s": 28,
+    "t": 7,
+    "u": 11,
+    "v": 21,
+    "z": 32,
+    "|": 4,
+    "á": 38,
+    "í": 36,
+    "ó": 35,
+    "ā": 17,
+    "ē": 25,
+    "ī": 23,
+    "ŋ": 18,
+    "ō": 20,
+    "ū": 22,
+    "—": 27,
+    "ꞌ": 9
+  },
+  "rav": {
+    "'": 51,
+    "-": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "ँ": 18,
+    "ं": 50,
+    "ः": 45,
+    "अ": 37,
+    "आ": 28,
+    "इ": 56,
+    "ई": 23,
+    "उ": 54,
+    "ऊ": 33,
+    "ए": 35,
+    "ऐ": 55,
+    "ओ": 49,
+    "क": 9,
+    "ख": 16,
+    "ग": 34,
+    "घ": 48,
+    "ङ": 26,
+    "च": 20,
+    "छ": 29,
+    "ज": 40,
+    "झ": 36,
+    "ट": 41,
+    "ठ": 46,
+    "ड": 44,
+    "ढ": 43,
+    "ण": 57,
+    "त": 14,
+    "थ": 39,
+    "द": 32,
+    "ध": 31,
+    "न": 10,
+    "प": 17,
+    "फ": 42,
+    "ब": 25,
+    "भ": 38,
+    "म": 7,
+    "य": 24,
+    "र": 22,
+    "ल": 13,
+    "व": 21,
+    "श": 58,
+    "स": 15,
+    "ह": 30,
+    "़": 47,
+    "ा": 4,
+    "ी": 6,
+    "ू": 11,
+    "ृ": 53,
+    "े": 27,
+    "ो": 19,
+    "्": 8,
+    "‍": 12
+  },
+  "raw": {
+    "'": 52,
+    "-": 49,
+    "0": 51,
+    "1": 53,
+    "2": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 35,
+    "c": 32,
+    "d": 16,
+    "e": 15,
+    "f": 50,
+    "g": 7,
+    "h": 14,
+    "i": 26,
+    "j": 44,
+    "k": 23,
+    "l": 17,
+    "m": 10,
+    "n": 5,
+    "o": 38,
+    "p": 27,
+    "q": 19,
+    "r": 13,
+    "s": 11,
+    "t": 12,
+    "u": 34,
+    "v": 6,
+    "w": 21,
+    "y": 30,
+    "z": 37,
+    "|": 4,
+    "à": 18,
+    "á": 31,
+    "è": 25,
+    "é": 40,
+    "ì": 22,
+    "í": 20,
+    "ï": 48,
+    "ò": 28,
+    "ó": 41,
+    "ø": 9,
+    "ù": 43,
+    "ú": 36,
+    "ā": 42,
+    "ē": 46,
+    "ō": 47,
+    "ū": 45,
+    "ǿ": 33,
+    "̀": 24,
+    "́": 29,
+    "̄": 39
+  },
+  "rej": {
+    "'": 24,
+    "-": 25,
+    "0": 30,
+    "1": 32,
+    "2": 34,
+    "4": 33,
+    "6": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 26,
+    "d": 16,
+    "e": 6,
+    "f": 28,
+    "g": 18,
+    "h": 21,
+    "i": 9,
+    "j": 22,
+    "k": 10,
+    "l": 17,
+    "m": 14,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "r": 20,
+    "s": 13,
+    "t": 12,
+    "u": 8,
+    "v": 31,
+    "w": 27,
+    "y": 23,
+    "z": 29,
+    "|": 4
+  },
+  "rel": {
+    "'": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 24,
+    "d": 16,
+    "e": 7,
+    "f": 26,
+    "g": 22,
+    "h": 9,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 18,
+    "m": 19,
+    "n": 10,
+    "o": 8,
+    "r": 15,
+    "s": 13,
+    "t": 11,
+    "u": 14,
+    "w": 20,
+    "y": 17,
+    "|": 4,
+    "á": 28,
+    "é": 30,
+    "í": 27,
+    "ó": 31,
+    "ú": 29
+  },
+  "rgu": {
+    "'": 16,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 17,
+    "e": 6,
+    "f": 21,
+    "g": 25,
+    "h": 19,
+    "i": 9,
+    "k": 20,
+    "l": 13,
+    "m": 11,
+    "n": 7,
+    "o": 10,
+    "p": 22,
+    "r": 14,
+    "s": 12,
+    "t": 15,
+    "u": 8,
+    "y": 24,
+    "|": 5
+  },
+  "rhg": {
+    "'": 36,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 29,
+    "d": 14,
+    "e": 10,
+    "f": 28,
+    "g": 24,
+    "h": 19,
+    "i": 6,
+    "j": 31,
+    "k": 21,
+    "l": 15,
+    "m": 25,
+    "n": 12,
+    "o": 8,
+    "p": 35,
+    "r": 7,
+    "s": 26,
+    "t": 9,
+    "u": 16,
+    "v": 37,
+    "w": 34,
+    "y": 23,
+    "z": 27,
+    "|": 4,
+    "á": 11,
+    "ç": 32,
+    "é": 18,
+    "í": 20,
+    "ñ": 22,
+    "ó": 13,
+    "ú": 30
+  },
+  "rif-script_arabic": {
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ء": 44,
+    "آ": 40,
+    "أ": 22,
+    "ؤ": 38,
+    "إ": 31,
+    "ئ": 36,
+    "ا": 5,
+    "ب": 16,
+    "ت": 15,
+    "ث": 13,
+    "ج": 23,
+    "ح": 24,
+    "خ": 29,
+    "د": 20,
+    "ذ": 12,
+    "ر": 10,
+    "ز": 28,
+    "س": 11,
+    "ش": 17,
+    "ص": 34,
+    "ض": 35,
+    "ط": 37,
+    "ظ": 42,
+    "ع": 21,
+    "غ": 18,
+    "ف": 30,
+    "ق": 25,
+    "ك": 27,
+    "ل": 19,
+    "م": 9,
+    "ن": 7,
+    "ه": 33,
+    "و": 8,
+    "ى": 39,
+    "ي": 6,
+    "پ": 43,
+    "ڒ": 14,
+    "ژ": 41,
+    "ݣ": 32
+  },
+  "rif-script_latin": {
+    "'": 43,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 20,
+    "d": 23,
+    "e": 8,
+    "f": 31,
+    "g": 33,
+    "h": 36,
+    "i": 7,
+    "j": 26,
+    "k": 28,
+    "l": 22,
+    "m": 9,
+    "n": 6,
+    "p": 41,
+    "q": 25,
+    "r": 10,
+    "s": 11,
+    "t": 18,
+    "u": 13,
+    "w": 14,
+    "x": 32,
+    "y": 16,
+    "z": 27,
+    "|": 4,
+    "č": 42,
+    "ƹ": 24,
+    "ǧ": 38,
+    "ȓ": 19,
+    "ɣ": 21,
+    "ʼ": 39,
+    "ḍ": 37,
+    "ḏ": 12,
+    "ḥ": 29,
+    "ṣ": 34,
+    "ṭ": 35,
+    "ṯ": 15,
+    "ẓ": 40
+  },
+  "ril": {
+    "'": 29,
+    "0": 28,
+    "1": 27,
+    "9": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 24,
+    "d": 22,
+    "e": 8,
+    "g": 19,
+    "h": 7,
+    "i": 12,
+    "k": 11,
+    "l": 20,
+    "m": 14,
+    "n": 9,
+    "o": 18,
+    "p": 17,
+    "r": 16,
+    "s": 21,
+    "t": 13,
+    "u": 10,
+    "w": 15,
+    "y": 23,
+    "z": 6,
+    "|": 4
+  },
+  "rim": {
+    "'": 28,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 29,
+    "d": 22,
+    "e": 8,
+    "f": 23,
+    "g": 12,
+    "h": 18,
+    "i": 11,
+    "j": 24,
+    "k": 16,
+    "l": 27,
+    "m": 10,
+    "n": 7,
+    "o": 13,
+    "p": 30,
+    "r": 20,
+    "s": 19,
+    "t": 15,
+    "u": 6,
+    "v": 21,
+    "w": 14,
+    "y": 17,
+    "z": 31,
+    "|": 4,
+    "ë": 9,
+    "ü": 26
+  },
+  "rjs": {
+    "'": 60,
+    "1": 44,
+    "2": 57,
+    "3": 58,
+    "4": 61,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "i": 45,
+    "l": 59,
+    "m": 50,
+    "q": 51,
+    "|": 4,
+    "ँ": 24,
+    "ं": 49,
+    "अ": 34,
+    "आ": 22,
+    "इ": 25,
+    "उ": 32,
+    "ए": 21,
+    "ओ": 55,
+    "क": 8,
+    "ख": 26,
+    "ग": 31,
+    "घ": 48,
+    "ङ": 41,
+    "च": 28,
+    "छ": 37,
+    "ज": 27,
+    "झ": 47,
+    "ट": 33,
+    "ठ": 43,
+    "ड": 29,
+    "ढ": 56,
+    "त": 12,
+    "थ": 42,
+    "द": 23,
+    "ध": 40,
+    "न": 10,
+    "ऩ": 52,
+    "प": 20,
+    "फ": 46,
+    "ब": 18,
+    "भ": 38,
+    "म": 15,
+    "य": 35,
+    "र": 6,
+    "ऱ": 54,
+    "ल": 13,
+    "व": 39,
+    "श": 64,
+    "स": 14,
+    "ह": 16,
+    "़": 36,
+    "ा": 5,
+    "ि": 9,
+    "ु": 17,
+    "ृ": 53,
+    "े": 7,
+    "ो": 30,
+    "्": 11,
+    "३": 63,
+    "६": 62,
+    "‍": 19
+  },
+  "rkt": {
+    "'": 61,
+    "-": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ঁ": 55,
+    "ং": 42,
+    "ঃ": 56,
+    "অ": 43,
+    "আ": 26,
+    "ই": 18,
+    "ঈ": 63,
+    "উ": 28,
+    "ঋ": 62,
+    "এ": 35,
+    "ঐ": 51,
+    "ও": 39,
+    "ঔ": 65,
+    "ক": 9,
+    "খ": 36,
+    "গ": 24,
+    "ঘ": 47,
+    "ঙ": 48,
+    "চ": 27,
+    "ছ": 37,
+    "জ": 31,
+    "ঝ": 54,
+    "ঞ": 52,
+    "ট": 30,
+    "ঠ": 46,
+    "ড": 41,
+    "ঢ": 59,
+    "ণ": 50,
+    "ত": 15,
+    "থ": 40,
+    "দ": 19,
+    "ধ": 38,
+    "ন": 11,
+    "প": 23,
+    "ফ": 44,
+    "ব": 17,
+    "ভ": 32,
+    "ম": 12,
+    "য": 10,
+    "র": 6,
+    "ল": 16,
+    "শ": 29,
+    "ষ": 33,
+    "স": 21,
+    "হ": 25,
+    "়": 14,
+    "া": 5,
+    "ি": 8,
+    "ী": 34,
+    "ু": 20,
+    "ূ": 45,
+    "ৃ": 60,
+    "ে": 7,
+    "ৈ": 57,
+    "ো": 22,
+    "ৌ": 53,
+    "্": 13,
+    "ৎ": 58,
+    "—": 64
+  },
+  "rmc-script_cyrillic": {
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 25,
+    "в": 15,
+    "г": 21,
+    "д": 12,
+    "е": 6,
+    "ж": 29,
+    "з": 30,
+    "и": 28,
+    "й": 23,
+    "к": 14,
+    "л": 11,
+    "м": 19,
+    "н": 7,
+    "о": 8,
+    "п": 17,
+    "р": 9,
+    "с": 10,
+    "т": 16,
+    "у": 20,
+    "ф": 36,
+    "х": 32,
+    "ц": 33,
+    "ч": 26,
+    "ш": 27,
+    "щ": 38,
+    "ь": 34,
+    "ю": 37,
+    "я": 22,
+    "є": 18,
+    "і": 13,
+    "ї": 35,
+    "ґ": 24,
+    "—": 31
+  },
+  "rmc-script_latin": {
+    "'": 36,
+    "-": 34,
+    "1": 40,
+    "2": 42,
+    "3": 43,
+    "5": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 39,
+    "a": 5,
+    "b": 22,
+    "c": 28,
+    "d": 13,
+    "e": 6,
+    "f": 32,
+    "g": 25,
+    "h": 15,
+    "i": 12,
+    "j": 21,
+    "k": 14,
+    "l": 11,
+    "m": 19,
+    "n": 8,
+    "o": 7,
+    "p": 18,
+    "r": 9,
+    "s": 10,
+    "t": 16,
+    "u": 20,
+    "v": 17,
+    "x": 37,
+    "y": 35,
+    "z": 31,
+    "|": 4,
+    "ó": 38,
+    "č": 24,
+    "ď": 30,
+    "ľ": 27,
+    "ň": 29,
+    "š": 23,
+    "ť": 33,
+    "ž": 26
+  },
+  "rmo": {
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 23,
+    "d": 16,
+    "e": 5,
+    "f": 25,
+    "g": 24,
+    "h": 15,
+    "i": 12,
+    "j": 21,
+    "k": 13,
+    "l": 9,
+    "m": 19,
+    "n": 7,
+    "o": 8,
+    "p": 18,
+    "r": 10,
+    "s": 11,
+    "t": 14,
+    "u": 17,
+    "v": 27,
+    "w": 20,
+    "x": 29,
+    "y": 30,
+    "z": 28,
+    "|": 4
+  },
+  "rmy-script_cyrillic": {
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "c": 38,
+    "|": 4,
+    "а": 5,
+    "б": 25,
+    "в": 15,
+    "г": 27,
+    "д": 14,
+    "е": 22,
+    "ж": 26,
+    "з": 31,
+    "и": 10,
+    "й": 24,
+    "к": 16,
+    "л": 12,
+    "м": 18,
+    "н": 7,
+    "о": 9,
+    "п": 19,
+    "р": 11,
+    "с": 8,
+    "т": 13,
+    "у": 17,
+    "ф": 34,
+    "х": 21,
+    "ц": 32,
+    "ч": 29,
+    "ш": 23,
+    "щ": 36,
+    "ы": 28,
+    "ь": 35,
+    "э": 6,
+    "ю": 37,
+    "я": 20,
+    "ё": 33,
+    "—": 30
+  },
+  "rmy-script_latin": {
+    "'": 29,
+    "8": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 22,
+    "d": 15,
+    "e": 6,
+    "f": 26,
+    "g": 20,
+    "h": 21,
+    "i": 7,
+    "j": 25,
+    "k": 11,
+    "l": 9,
+    "m": 18,
+    "n": 10,
+    "o": 8,
+    "p": 16,
+    "r": 13,
+    "s": 14,
+    "t": 12,
+    "u": 17,
+    "v": 19,
+    "z": 24,
+    "|": 4,
+    "ñ": 28,
+    "—": 27
+  },
+  "rng": {
+    "'": 25,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 36,
+    "d": 29,
+    "e": 10,
+    "f": 24,
+    "g": 20,
+    "h": 14,
+    "i": 6,
+    "k": 9,
+    "l": 11,
+    "m": 13,
+    "n": 7,
+    "o": 16,
+    "p": 28,
+    "r": 31,
+    "s": 19,
+    "t": 17,
+    "u": 8,
+    "v": 35,
+    "w": 15,
+    "y": 18,
+    "z": 34,
+    "|": 4,
+    "ï": 42,
+    "ŝ": 22,
+    "š": 23,
+    "ž": 37,
+    "ḇ": 33,
+    "ḍ": 21,
+    "ḳ": 45,
+    "ṁ": 40,
+    "ṅ": 27,
+    "ṛ": 26,
+    "ṡ": 38,
+    "ṣ": 41,
+    "ṭ": 30,
+    "ẑ": 44,
+    "ẹ": 39,
+    "ị": 46,
+    "—": 43
+  },
+  "rnl": {
+    "'": 36,
+    "-": 31,
+    "0": 32,
+    "1": 33,
+    "2": 34,
+    "3": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 20,
+    "d": 23,
+    "e": 14,
+    "f": 38,
+    "g": 12,
+    "h": 8,
+    "i": 7,
+    "j": 30,
+    "k": 11,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 19,
+    "p": 21,
+    "q": 35,
+    "r": 15,
+    "s": 18,
+    "t": 10,
+    "u": 9,
+    "v": 24,
+    "z": 26,
+    "|": 4,
+    "ā": 16,
+    "ē": 25,
+    "ī": 29,
+    "ō": 22,
+    "ū": 28
+  },
+  "roh-dialect_sursilv": {
+    "'": 26,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 14,
+    "d": 15,
+    "e": 6,
+    "f": 22,
+    "g": 17,
+    "h": 21,
+    "i": 7,
+    "j": 27,
+    "k": 31,
+    "l": 10,
+    "m": 19,
+    "n": 9,
+    "o": 16,
+    "p": 18,
+    "q": 25,
+    "r": 12,
+    "s": 8,
+    "t": 13,
+    "u": 11,
+    "v": 20,
+    "w": 34,
+    "x": 28,
+    "y": 35,
+    "z": 23,
+    "|": 4,
+    "à": 36,
+    "ä": 37,
+    "è": 33,
+    "é": 29,
+    "ò": 39,
+    "ü": 38,
+    "–": 32
+  },
+  "roh-dialect_vallader": {
+    "'": 26,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 15,
+    "d": 14,
+    "e": 9,
+    "f": 24,
+    "g": 20,
+    "h": 19,
+    "i": 7,
+    "j": 29,
+    "k": 34,
+    "l": 11,
+    "m": 18,
+    "n": 8,
+    "o": 16,
+    "p": 17,
+    "q": 27,
+    "r": 10,
+    "s": 6,
+    "t": 12,
+    "u": 13,
+    "v": 21,
+    "w": 36,
+    "x": 31,
+    "y": 38,
+    "z": 23,
+    "|": 4,
+    "à": 28,
+    "â": 39,
+    "ä": 40,
+    "ç": 41,
+    "è": 35,
+    "ê": 42,
+    "ò": 32,
+    "ö": 30,
+    "ü": 22,
+    "–": 37
+  },
+  "rol": {
+    "'": 37,
+    "-": 26,
+    "0": 30,
+    "1": 34,
+    "2": 36,
+    "3": 45,
+    "4": 35,
+    "5": 39,
+    "6": 46,
+    "7": 44,
+    "8": 47,
+    "9": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 29,
+    "d": 16,
+    "e": 22,
+    "f": 31,
+    "g": 7,
+    "h": 19,
+    "i": 8,
+    "j": 28,
+    "k": 11,
+    "l": 20,
+    "m": 12,
+    "n": 6,
+    "o": 9,
+    "p": 17,
+    "q": 40,
+    "r": 21,
+    "s": 10,
+    "t": 13,
+    "u": 15,
+    "v": 33,
+    "w": 23,
+    "x": 38,
+    "y": 14,
+    "z": 32,
+    "|": 5,
+    "à": 24,
+    "á": 41,
+    "è": 48,
+    "ì": 25,
+    "ñ": 42,
+    "ò": 27
+  },
+  "ron": {
+    "!": 59,
+    "$": 77,
+    "%": 53,
+    "'": 54,
+    "+": 73,
+    ",": 52,
+    "-": 30,
+    ".": 36,
+    "/": 51,
+    "0": 32,
+    "1": 35,
+    "2": 37,
+    "3": 45,
+    "4": 42,
+    "5": 41,
+    "6": 47,
+    "7": 48,
+    "8": 46,
+    "9": 40,
+    ":": 49,
+    ";": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 69,
+    "[": 71,
+    "]": 72,
+    "a": 6,
+    "b": 25,
+    "c": 13,
+    "d": 16,
+    "e": 5,
+    "f": 20,
+    "g": 24,
+    "h": 29,
+    "i": 7,
+    "j": 31,
+    "k": 34,
+    "l": 12,
+    "m": 19,
+    "n": 10,
+    "o": 14,
+    "p": 17,
+    "q": 50,
+    "r": 8,
+    "s": 15,
+    "t": 9,
+    "u": 11,
+    "v": 21,
+    "w": 39,
+    "x": 33,
+    "y": 38,
+    "z": 27,
+    "|": 4,
+    "¥": 64,
+    "«": 43,
+    "°": 68,
+    "²": 66,
+    "»": 44,
+    "á": 56,
+    "â": 28,
+    "ã": 74,
+    "ç": 67,
+    "é": 65,
+    "í": 61,
+    "î": 23,
+    "õ": 62,
+    "ö": 78,
+    "ú": 63,
+    "ü": 57,
+    "ă": 18,
+    "ć": 82,
+    "đ": 81,
+    "ł": 76,
+    "ş": 80,
+    "ţ": 79,
+    "ș": 26,
+    "ț": 22,
+    "̇": 75,
+    "–": 60,
+    "—": 70,
+    "”": 58
+  },
+  "rop": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "d": 12,
+    "e": 8,
+    "f": 25,
+    "g": 15,
+    "h": 24,
+    "i": 6,
+    "j": 20,
+    "k": 21,
+    "l": 10,
+    "m": 11,
+    "n": 7,
+    "o": 14,
+    "p": 22,
+    "r": 16,
+    "s": 18,
+    "t": 13,
+    "u": 17,
+    "w": 19,
+    "y": 23,
+    "|": 4
+  },
+  "rro": {
+    "'": 16,
+    "0": 29,
+    "3": 28,
+    "4": 32,
+    "5": 31,
+    "9": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 13,
+    "d": 22,
+    "e": 8,
+    "f": 26,
+    "g": 24,
+    "h": 11,
+    "i": 6,
+    "k": 10,
+    "l": 21,
+    "m": 15,
+    "n": 7,
+    "o": 9,
+    "p": 18,
+    "r": 14,
+    "s": 20,
+    "t": 17,
+    "u": 12,
+    "v": 27,
+    "w": 19,
+    "y": 30,
+    "|": 5,
+    "ṉ": 25,
+    "ṯ": 23,
+    "—": 33
+  },
+  "rub": {
+    "'": 34,
+    "-": 31,
+    "0": 38,
+    "3": 37,
+    "4": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 8,
+    "c": 26,
+    "d": 23,
+    "e": 10,
+    "f": 29,
+    "g": 18,
+    "h": 21,
+    "i": 6,
+    "j": 28,
+    "k": 12,
+    "l": 20,
+    "m": 13,
+    "n": 11,
+    "o": 14,
+    "p": 27,
+    "r": 17,
+    "s": 22,
+    "t": 19,
+    "u": 7,
+    "v": 32,
+    "w": 16,
+    "y": 15,
+    "z": 25,
+    "|": 5,
+    "ŋ": 33,
+    "ʼ": 30,
+    "̯": 24,
+    "̱": 9,
+    "—": 35
+  },
+  "ruf": {
+    "'": 25,
+    "-": 30,
+    "0": 31,
+    "1": 32,
+    "2": 33,
+    "3": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 24,
+    "d": 21,
+    "e": 7,
+    "f": 23,
+    "g": 14,
+    "h": 15,
+    "i": 6,
+    "j": 28,
+    "k": 12,
+    "l": 10,
+    "m": 13,
+    "n": 8,
+    "o": 16,
+    "p": 22,
+    "r": 26,
+    "s": 18,
+    "t": 17,
+    "u": 9,
+    "v": 29,
+    "w": 11,
+    "y": 19,
+    "z": 27,
+    "|": 5
+  },
+  "rug": {
+    "'": 26,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 20,
+    "e": 7,
+    "g": 18,
+    "h": 22,
+    "i": 6,
+    "k": 13,
+    "l": 17,
+    "m": 15,
+    "n": 11,
+    "o": 10,
+    "p": 16,
+    "q": 25,
+    "r": 14,
+    "s": 8,
+    "t": 12,
+    "u": 9,
+    "v": 19,
+    "z": 24,
+    "|": 4,
+    "ṉ": 23
+  },
+  "run": {
+    "'": 22,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 25,
+    "d": 23,
+    "e": 9,
+    "f": 28,
+    "g": 17,
+    "h": 20,
+    "i": 6,
+    "j": 26,
+    "k": 14,
+    "l": 29,
+    "m": 13,
+    "n": 8,
+    "o": 11,
+    "p": 27,
+    "r": 10,
+    "s": 19,
+    "t": 18,
+    "u": 7,
+    "v": 24,
+    "w": 15,
+    "y": 16,
+    "z": 21,
+    "|": 4,
+    "—": 31
+  },
+  "rus": {
+    "!": 92,
+    "\"": 78,
+    "%": 81,
+    "&": 90,
+    "'": 82,
+    "+": 87,
+    ",": 76,
+    "-": 37,
+    ".": 63,
+    "/": 71,
+    "0": 36,
+    "1": 39,
+    "2": 41,
+    "3": 51,
+    "4": 52,
+    "5": 49,
+    "6": 53,
+    "7": 60,
+    "8": 55,
+    "9": 47,
+    ":": 69,
+    ";": 86,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 88,
+    "]": 89,
+    "a": 40,
+    "b": 67,
+    "c": 57,
+    "d": 61,
+    "e": 42,
+    "f": 75,
+    "g": 62,
+    "h": 58,
+    "i": 44,
+    "j": 77,
+    "k": 66,
+    "l": 54,
+    "m": 65,
+    "n": 45,
+    "o": 43,
+    "p": 59,
+    "q": 83,
+    "r": 46,
+    "s": 50,
+    "t": 48,
+    "u": 64,
+    "v": 68,
+    "w": 73,
+    "x": 74,
+    "y": 72,
+    "z": 79,
+    "|": 4,
+    "°": 91,
+    "õ": 85,
+    "́": 93,
+    "а": 8,
+    "б": 25,
+    "в": 13,
+    "г": 22,
+    "д": 17,
+    "е": 6,
+    "ж": 29,
+    "з": 24,
+    "и": 7,
+    "й": 27,
+    "к": 15,
+    "л": 14,
+    "м": 16,
+    "н": 9,
+    "о": 5,
+    "п": 18,
+    "р": 12,
+    "с": 11,
+    "т": 10,
+    "у": 19,
+    "ф": 35,
+    "х": 28,
+    "ц": 32,
+    "ч": 26,
+    "ш": 31,
+    "щ": 33,
+    "ъ": 56,
+    "ы": 20,
+    "ь": 23,
+    "э": 34,
+    "ю": 30,
+    "я": 21,
+    "ё": 38,
+    "‐": 94,
+    "–": 84,
+    "—": 70,
+    "№": 80
+  },
+  "sab": {
+    "'": 39,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 17,
+    "d": 19,
+    "e": 6,
+    "f": 29,
+    "g": 8,
+    "h": 18,
+    "i": 7,
+    "j": 23,
+    "k": 14,
+    "l": 9,
+    "m": 20,
+    "n": 11,
+    "o": 15,
+    "p": 25,
+    "q": 35,
+    "r": 16,
+    "s": 22,
+    "t": 13,
+    "u": 10,
+    "v": 30,
+    "w": 21,
+    "x": 38,
+    "y": 36,
+    "z": 34,
+    "|": 4,
+    "á": 31,
+    "é": 26,
+    "í": 28,
+    "ñ": 24,
+    "ó": 33,
+    "ú": 37,
+    "—": 27
+  },
+  "sag": {
+    "'": 38,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 32,
+    "d": 22,
+    "e": 7,
+    "f": 25,
+    "g": 12,
+    "h": 26,
+    "i": 6,
+    "j": 29,
+    "k": 14,
+    "l": 11,
+    "m": 16,
+    "n": 8,
+    "o": 9,
+    "p": 20,
+    "q": 39,
+    "r": 21,
+    "s": 13,
+    "t": 10,
+    "u": 18,
+    "v": 33,
+    "w": 24,
+    "x": 40,
+    "y": 17,
+    "z": 19,
+    "|": 4,
+    "â": 27,
+    "è": 41,
+    "é": 30,
+    "ê": 34,
+    "ë": 23,
+    "î": 42,
+    "ï": 35,
+    "ô": 36,
+    "ö": 31,
+    "–": 37
+  },
+  "sah": {
+    "-": 31,
+    "3": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 45,
+    "c": 46,
+    "h": 47,
+    "i": 48,
+    "o": 40,
+    "y": 42,
+    "|": 4,
+    "а": 5,
+    "б": 15,
+    "в": 33,
+    "г": 25,
+    "д": 18,
+    "е": 32,
+    "ж": 39,
+    "з": 35,
+    "и": 9,
+    "й": 23,
+    "к": 16,
+    "л": 14,
+    "м": 24,
+    "н": 8,
+    "о": 13,
+    "п": 27,
+    "р": 10,
+    "с": 17,
+    "т": 7,
+    "у": 12,
+    "ф": 36,
+    "х": 19,
+    "ц": 37,
+    "ч": 30,
+    "ш": 38,
+    "щ": 41,
+    "ы": 11,
+    "ь": 28,
+    "э": 6,
+    "ю": 44,
+    "я": 34,
+    "ҕ": 26,
+    "ҥ": 29,
+    "ү": 20,
+    "һ": 22,
+    "ө": 21,
+    "–": 43
+  },
+  "saj": {
+    "'": 10,
+    "-": 19,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 30,
+    "d": 18,
+    "e": 12,
+    "f": 28,
+    "g": 9,
+    "h": 27,
+    "i": 8,
+    "j": 24,
+    "k": 23,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 7,
+    "p": 26,
+    "r": 15,
+    "s": 14,
+    "t": 16,
+    "u": 11,
+    "w": 22,
+    "y": 25,
+    "z": 31,
+    "|": 5,
+    "̱": 32,
+    "ḇ": 29,
+    "ḏ": 20
+  },
+  "saq": {
+    "'": 22,
+    "-": 34,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "3": 37,
+    "4": 29,
+    "5": 32,
+    "6": 33,
+    "7": 36,
+    "8": 35,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 21,
+    "d": 25,
+    "e": 8,
+    "g": 18,
+    "h": 20,
+    "i": 6,
+    "j": 24,
+    "k": 10,
+    "l": 13,
+    "m": 16,
+    "n": 7,
+    "o": 9,
+    "p": 17,
+    "r": 14,
+    "s": 19,
+    "t": 11,
+    "u": 12,
+    "w": 26,
+    "y": 15,
+    "|": 4,
+    "꞉": 27
+  },
+  "sas": {
+    "'": 29,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 28,
+    "d": 10,
+    "e": 6,
+    "f": 31,
+    "g": 12,
+    "h": 22,
+    "i": 8,
+    "j": 24,
+    "k": 17,
+    "l": 14,
+    "m": 16,
+    "n": 7,
+    "o": 23,
+    "p": 18,
+    "q": 13,
+    "r": 19,
+    "s": 9,
+    "t": 11,
+    "u": 15,
+    "w": 27,
+    "y": 26,
+    "z": 30,
+    "|": 4,
+    "é": 32,
+    "ẽ": 20
+  },
+  "sat": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ᱚ": 14,
+    "ᱛ": 17,
+    "ᱜ": 12,
+    "ᱝ": 36,
+    "ᱞ": 19,
+    "ᱟ": 5,
+    "ᱠ": 8,
+    "ᱡ": 30,
+    "ᱢ": 13,
+    "ᱣ": 26,
+    "ᱤ": 10,
+    "ᱥ": 22,
+    "ᱦ": 28,
+    "ᱧ": 18,
+    "ᱨ": 20,
+    "ᱩ": 6,
+    "ᱪ": 31,
+    "ᱫ": 15,
+    "ᱬ": 37,
+    "ᱭ": 11,
+    "ᱮ": 9,
+    "ᱯ": 33,
+    "ᱰ": 29,
+    "ᱱ": 7,
+    "ᱲ": 32,
+    "ᱴ": 21,
+    "ᱵ": 24,
+    "ᱶ": 35,
+    "ᱷ": 23,
+    "ᱸ": 34,
+    "ᱹ": 27,
+    "ᱼ": 25,
+    "ᱽ": 38,
+    "᱾": 16
+  },
+  "sba": {
+    "'": 22,
+    "-": 31,
+    "1": 49,
+    "2": 48,
+    "3": 52,
+    "4": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "d": 10,
+    "e": 6,
+    "g": 8,
+    "h": 43,
+    "i": 15,
+    "j": 16,
+    "k": 17,
+    "l": 11,
+    "m": 12,
+    "n": 9,
+    "o": 14,
+    "p": 35,
+    "r": 13,
+    "s": 21,
+    "t": 18,
+    "u": 25,
+    "v": 50,
+    "w": 26,
+    "y": 28,
+    "|": 4,
+    "à": 40,
+    "á": 34,
+    "è": 37,
+    "é": 20,
+    "í": 33,
+    "ò": 41,
+    "ó": 39,
+    "ú": 44,
+    "ŋ": 29,
+    "ǝ": 47,
+    "ɓ": 30,
+    "ɔ": 19,
+    "ɗ": 42,
+    "ə": 7,
+    "ɛ": 38,
+    "̀": 46,
+    "́": 23,
+    "̰": 32,
+    "ḛ": 27,
+    "ḭ": 36,
+    "ṵ": 45
+  },
+  "sbd": {
+    "-": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 15,
+    "e": 17,
+    "f": 35,
+    "g": 18,
+    "h": 41,
+    "i": 8,
+    "k": 11,
+    "l": 9,
+    "m": 14,
+    "n": 6,
+    "o": 12,
+    "p": 27,
+    "r": 20,
+    "s": 19,
+    "t": 23,
+    "u": 21,
+    "v": 48,
+    "w": 16,
+    "y": 24,
+    "z": 28,
+    "|": 4,
+    "à": 31,
+    "á": 22,
+    "ã": 29,
+    "è": 49,
+    "é": 50,
+    "ì": 39,
+    "í": 36,
+    "ò": 42,
+    "ó": 51,
+    "ù": 44,
+    "ú": 45,
+    "ĩ": 37,
+    "ń": 47,
+    "ũ": 40,
+    "ǎ": 38,
+    "ǐ": 43,
+    "ǹ": 26,
+    "ɔ": 10,
+    "ə": 30,
+    "ɛ": 7,
+    "̀": 34,
+    "́": 25,
+    "̃": 32,
+    "̌": 52,
+    "‐": 33,
+    "—": 46
+  },
+  "sbl": {
+    "'": 29,
+    "-": 17,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "d": 23,
+    "e": 22,
+    "g": 14,
+    "h": 11,
+    "i": 7,
+    "k": 13,
+    "l": 15,
+    "m": 10,
+    "n": 6,
+    "o": 8,
+    "p": 16,
+    "r": 21,
+    "s": 20,
+    "t": 12,
+    "w": 19,
+    "y": 9,
+    "|": 5,
+    "à": 24,
+    "è": 26,
+    "ì": 30,
+    "ò": 25,
+    "̀": 28,
+    "̇": 27,
+    "—": 31
+  },
+  "sbp": {
+    "'": 28,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "d": 22,
+    "e": 10,
+    "f": 24,
+    "g": 19,
+    "h": 17,
+    "i": 6,
+    "j": 20,
+    "k": 11,
+    "l": 9,
+    "m": 14,
+    "n": 8,
+    "o": 12,
+    "p": 23,
+    "s": 16,
+    "t": 21,
+    "u": 7,
+    "v": 15,
+    "w": 13,
+    "y": 18,
+    "|": 5,
+    "á": 26,
+    "ʼ": 27
+  },
+  "sch": {
+    "'": 31,
+    "-": 32,
+    "0": 38,
+    "1": 41,
+    "4": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 26,
+    "c": 21,
+    "d": 23,
+    "e": 9,
+    "f": 34,
+    "g": 11,
+    "h": 8,
+    "i": 7,
+    "j": 25,
+    "k": 15,
+    "l": 17,
+    "m": 10,
+    "n": 5,
+    "o": 14,
+    "p": 19,
+    "q": 39,
+    "r": 16,
+    "s": 18,
+    "t": 12,
+    "u": 13,
+    "v": 30,
+    "w": 35,
+    "x": 36,
+    "y": 33,
+    "z": 28,
+    "|": 4,
+    "â": 20,
+    "ê": 24,
+    "î": 29,
+    "ô": 22,
+    "û": 27,
+    "—": 37
+  },
+  "sck": {
+    "'": 59,
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 19,
+    "ः": 57,
+    "अ": 43,
+    "आ": 20,
+    "इ": 27,
+    "ई": 35,
+    "उ": 14,
+    "ऊ": 55,
+    "ए": 41,
+    "ऐ": 62,
+    "ओ": 53,
+    "क": 7,
+    "ख": 34,
+    "ग": 25,
+    "घ": 49,
+    "च": 33,
+    "छ": 47,
+    "ज": 24,
+    "झ": 48,
+    "ञ": 56,
+    "ट": 40,
+    "ठ": 45,
+    "ड": 37,
+    "ढ": 51,
+    "ण": 54,
+    "त": 17,
+    "थ": 44,
+    "द": 29,
+    "ध": 38,
+    "न": 9,
+    "प": 22,
+    "फ": 50,
+    "ब": 18,
+    "भ": 36,
+    "म": 11,
+    "य": 21,
+    "र": 6,
+    "ल": 12,
+    "व": 30,
+    "श": 32,
+    "ष": 52,
+    "स": 15,
+    "ह": 10,
+    "़": 39,
+    "ा": 8,
+    "ि": 16,
+    "ी": 23,
+    "ु": 26,
+    "ू": 42,
+    "ृ": 61,
+    "े": 5,
+    "ै": 60,
+    "ो": 13,
+    "ौ": 58,
+    "्": 28,
+    "‍": 31
+  },
+  "sda": {
+    "'": 21,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 29,
+    "d": 18,
+    "e": 14,
+    "f": 25,
+    "g": 17,
+    "h": 24,
+    "i": 7,
+    "j": 30,
+    "k": 12,
+    "l": 15,
+    "m": 10,
+    "n": 6,
+    "o": 11,
+    "p": 16,
+    "r": 19,
+    "s": 13,
+    "t": 9,
+    "u": 8,
+    "w": 28,
+    "y": 22,
+    "z": 27,
+    "|": 5,
+    "ë": 31,
+    "–": 26
+  },
+  "sea": {
+    "'": 27,
+    "-": 31,
+    "0": 38,
+    "1": 35,
+    "2": 39,
+    "4": 36,
+    "5": 37,
+    "6": 41,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 28,
+    "d": 24,
+    "e": 6,
+    "f": 33,
+    "g": 18,
+    "h": 11,
+    "i": 9,
+    "j": 23,
+    "k": 10,
+    "l": 16,
+    "m": 20,
+    "n": 7,
+    "o": 26,
+    "p": 19,
+    "q": 12,
+    "r": 13,
+    "s": 15,
+    "t": 17,
+    "u": 8,
+    "w": 30,
+    "y": 14,
+    "z": 34,
+    "|": 4,
+    "è": 25,
+    "é": 29,
+    "ñ": 32,
+    "ò": 21
+  },
+  "seh": {
+    "'": 27,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 25,
+    "d": 16,
+    "e": 9,
+    "f": 26,
+    "g": 23,
+    "h": 18,
+    "i": 7,
+    "j": 28,
+    "k": 10,
+    "l": 20,
+    "m": 11,
+    "n": 6,
+    "o": 12,
+    "p": 15,
+    "r": 24,
+    "s": 22,
+    "t": 14,
+    "u": 8,
+    "v": 29,
+    "w": 13,
+    "x": 30,
+    "y": 19,
+    "z": 21,
+    "|": 5
+  },
+  "ses": {
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 27,
+    "d": 14,
+    "e": 9,
+    "f": 26,
+    "g": 13,
+    "h": 19,
+    "i": 8,
+    "j": 28,
+    "k": 11,
+    "l": 20,
+    "m": 15,
+    "n": 7,
+    "o": 6,
+    "p": 32,
+    "r": 10,
+    "s": 21,
+    "t": 18,
+    "u": 16,
+    "w": 23,
+    "y": 12,
+    "z": 25,
+    "|": 4,
+    "ã": 35,
+    "õ": 34,
+    "ŋ": 22,
+    "š": 29,
+    "ũ": 38,
+    "ž": 31,
+    "ɲ": 30,
+    "ʼ": 24,
+    "ẽ": 36,
+    "–": 37
+  },
+  "sey": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 14,
+    "d": 25,
+    "e": 7,
+    "f": 32,
+    "g": 31,
+    "h": 20,
+    "i": 6,
+    "j": 12,
+    "k": 39,
+    "l": 26,
+    "m": 22,
+    "n": 19,
+    "o": 11,
+    "p": 13,
+    "q": 23,
+    "r": 18,
+    "s": 16,
+    "t": 21,
+    "u": 10,
+    "v": 36,
+    "x": 38,
+    "y": 17,
+    "z": 33,
+    "|": 4,
+    "á": 35,
+    "é": 30,
+    "ë": 9,
+    "í": 29,
+    "ñ": 24,
+    "ó": 34,
+    "ú": 28,
+    "̱": 15,
+    "̲": 37
+  },
+  "sgb": {
+    "-": 25,
+    "0": 35,
+    "1": 37,
+    "2": 38,
+    "3": 44,
+    "4": 43,
+    "5": 41,
+    "6": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 27,
+    "d": 23,
+    "e": 24,
+    "f": 30,
+    "g": 16,
+    "h": 10,
+    "i": 7,
+    "j": 26,
+    "k": 13,
+    "l": 12,
+    "m": 11,
+    "n": 6,
+    "o": 9,
+    "p": 15,
+    "q": 33,
+    "r": 20,
+    "s": 21,
+    "t": 14,
+    "u": 19,
+    "v": 32,
+    "w": 22,
+    "x": 36,
+    "y": 8,
+    "z": 31,
+    "|": 5,
+    "á": 28,
+    "ê": 17,
+    "í": 29,
+    "ñ": 40,
+    "ó": 39,
+    "ú": 34
+  },
+  "sgj": {
+    "'": 51,
+    "-": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 15,
+    "अ": 24,
+    "आ": 35,
+    "इ": 50,
+    "ई": 20,
+    "उ": 47,
+    "ऊ": 30,
+    "ए": 17,
+    "ओ": 26,
+    "क": 8,
+    "ख": 33,
+    "ग": 22,
+    "घ": 39,
+    "च": 32,
+    "छ": 45,
+    "ज": 21,
+    "झ": 43,
+    "ट": 37,
+    "ठ": 42,
+    "ड": 38,
+    "ढ": 48,
+    "त": 16,
+    "थ": 40,
+    "द": 27,
+    "ध": 44,
+    "न": 9,
+    "प": 23,
+    "फ": 46,
+    "ब": 18,
+    "भ": 36,
+    "म": 11,
+    "य": 34,
+    "र": 5,
+    "ल": 14,
+    "व": 31,
+    "स": 12,
+    "ह": 10,
+    "़": 41,
+    "ा": 6,
+    "ि": 25,
+    "ी": 13,
+    "ु": 29,
+    "ू": 28,
+    "े": 7,
+    "ो": 19,
+    "्": 52
+  },
+  "sgw": {
+    "'": 92,
+    "2": 243,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 242,
+    "|": 4,
+    "ለ": 85,
+    "ሉ": 208,
+    "ሊ": 148,
+    "ላ": 101,
+    "ሌ": 108,
+    "ል": 83,
+    "ሎ": 88,
+    "መ": 24,
+    "ሙ": 138,
+    "ሚ": 153,
+    "ማ": 27,
+    "ሜ": 67,
+    "ም": 5,
+    "ሞ": 87,
+    "ሟ": 102,
+    "ረ": 10,
+    "ሩ": 105,
+    "ሪ": 76,
+    "ራ": 50,
+    "ሬ": 60,
+    "ር": 6,
+    "ሮ": 31,
+    "ሰ": 22,
+    "ሱ": 63,
+    "ሲ": 137,
+    "ሳ": 59,
+    "ሴ": 115,
+    "ስ": 12,
+    "ሶ": 97,
+    "ሸ": 62,
+    "ሹ": 210,
+    "ሺ": 174,
+    "ሻ": 152,
+    "ሼ": 140,
+    "ሽ": 65,
+    "ሾ": 173,
+    "ቀ": 70,
+    "ቁ": 196,
+    "ቂ": 193,
+    "ቃ": 38,
+    "ቄ": 147,
+    "ቅ": 41,
+    "ቆ": 124,
+    "ቈ": 100,
+    "ቊ": 222,
+    "ቋ": 130,
+    "ቌ": 195,
+    "በ": 13,
+    "ቡ": 183,
+    "ቢ": 54,
+    "ባ": 23,
+    "ቤ": 68,
+    "ብ": 16,
+    "ቦ": 53,
+    "ቧ": 110,
+    "ተ": 14,
+    "ቱ": 131,
+    "ቲ": 64,
+    "ታ": 19,
+    "ቴ": 46,
+    "ት": 8,
+    "ቶ": 56,
+    "ቸ": 51,
+    "ቹ": 207,
+    "ቺ": 189,
+    "ቻ": 168,
+    "ቼ": 187,
+    "ች": 95,
+    "ቾ": 84,
+    "ነ": 11,
+    "ኑ": 145,
+    "ኒ": 111,
+    "ና": 26,
+    "ኔ": 73,
+    "ን": 7,
+    "ኖ": 35,
+    "ኛ": 165,
+    "ኜ": 246,
+    "ኝ": 229,
+    "አ": 15,
+    "ኡ": 191,
+    "ኢ": 61,
+    "ኤ": 49,
+    "እ": 21,
+    "ኦ": 109,
+    "ኧ": 44,
+    "ከ": 42,
+    "ኩ": 114,
+    "ኪ": 184,
+    "ካ": 89,
+    "ኬ": 180,
+    "ክ": 40,
+    "ኮ": 158,
+    "ኰ": 185,
+    "ኲ": 237,
+    "ኳ": 177,
+    "ኴ": 232,
+    "ኸ": 18,
+    "ኹ": 33,
+    "ኺ": 206,
+    "ኻ": 80,
+    "ኼ": 78,
+    "ኽ": 25,
+    "ኾ": 157,
+    "ዀ": 28,
+    "ዂ": 223,
+    "ዃ": 125,
+    "ዄ": 120,
+    "ወ": 32,
+    "ዉ": 162,
+    "ዊ": 71,
+    "ዋ": 91,
+    "ዌ": 57,
+    "ው": 48,
+    "ዎ": 182,
+    "ዕ": 227,
+    "ዘ": 29,
+    "ዙ": 161,
+    "ዚ": 218,
+    "ዛ": 118,
+    "ዜ": 160,
+    "ዝ": 34,
+    "ዞ": 129,
+    "ዠ": 127,
+    "ዢ": 216,
+    "ዣ": 236,
+    "ዤ": 200,
+    "ዥ": 139,
+    "ዦ": 205,
+    "የ": 9,
+    "ዩ": 134,
+    "ዪ": 228,
+    "ያ": 20,
+    "ዬ": 106,
+    "ይ": 17,
+    "ዮ": 75,
+    "ደ": 36,
+    "ዱ": 116,
+    "ዲ": 167,
+    "ዳ": 55,
+    "ዴ": 119,
+    "ድ": 43,
+    "ዶ": 79,
+    "ዼ": 240,
+    "ጀ": 107,
+    "ጁ": 213,
+    "ጂ": 164,
+    "ጃ": 163,
+    "ጄ": 194,
+    "ጅ": 98,
+    "ጆ": 247,
+    "ገ": 37,
+    "ጉ": 214,
+    "ጊ": 201,
+    "ጋ": 39,
+    "ጌ": 128,
+    "ግ": 30,
+    "ጎ": 150,
+    "ጐ": 81,
+    "ጒ": 224,
+    "ጓ": 170,
+    "ጔ": 154,
+    "ጠ": 47,
+    "ጡ": 203,
+    "ጢ": 99,
+    "ጣ": 74,
+    "ጤ": 121,
+    "ጥ": 45,
+    "ጦ": 96,
+    "ጨ": 90,
+    "ጩ": 245,
+    "ጪ": 171,
+    "ጫ": 113,
+    "ጬ": 142,
+    "ጭ": 82,
+    "ጮ": 169,
+    "ጰ": 225,
+    "ጱ": 239,
+    "ጲ": 186,
+    "ጳ": 126,
+    "ጴ": 143,
+    "ጵ": 175,
+    "ጶ": 190,
+    "ጹ": 241,
+    "ጻ": 244,
+    "ጼ": 238,
+    "ጽ": 235,
+    "ፀ": 146,
+    "ፃ": 176,
+    "ፅ": 155,
+    "ፆ": 198,
+    "ፈ": 58,
+    "ፉ": 234,
+    "ፊ": 159,
+    "ፋ": 93,
+    "ፌ": 156,
+    "ፍ": 52,
+    "ፎ": 122,
+    "ፏ": 188,
+    "ፐ": 72,
+    "ፑ": 221,
+    "ፒ": 204,
+    "ፓ": 136,
+    "ፔ": 117,
+    "ፕ": 144,
+    "ፖ": 149,
+    "ፗ": 211,
+    "ᎀ": 66,
+    "ᎁ": 212,
+    "ᎂ": 172,
+    "ᎃ": 94,
+    "ᎄ": 181,
+    "ᎅ": 226,
+    "ᎆ": 233,
+    "ᎇ": 151,
+    "ᎈ": 112,
+    "ᎉ": 166,
+    "ᎊ": 192,
+    "ᎋ": 132,
+    "ᎌ": 104,
+    "ᎍ": 220,
+    "ᎎ": 202,
+    "ᎏ": 209,
+    "ⷀ": 69,
+    "ⷃ": 197,
+    "ⷄ": 179,
+    "ⷅ": 103,
+    "ⷈ": 123,
+    "ⷋ": 215,
+    "ⷌ": 199,
+    "ⷍ": 135,
+    "ⷐ": 141,
+    "ⷒ": 230,
+    "ⷓ": 231,
+    "ⷔ": 217,
+    "ⷕ": 77,
+    "ⷘ": 86,
+    "ⷛ": 133,
+    "ⷝ": 178,
+    "ⷞ": 219
+  },
+  "shi": {
+    "1": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "d": 14,
+    "e": 35,
+    "f": 22,
+    "g": 21,
+    "h": 31,
+    "i": 6,
+    "j": 30,
+    "k": 17,
+    "l": 9,
+    "m": 15,
+    "n": 8,
+    "q": 26,
+    "r": 11,
+    "s": 12,
+    "t": 10,
+    "u": 13,
+    "w": 19,
+    "y": 18,
+    "z": 23,
+    "|": 4,
+    "â": 37,
+    "ÿ": 36,
+    "ġ": 16,
+    "š": 24,
+    "ε": 28,
+    "ḍ": 27,
+    "ḥ": 25,
+    "ḫ": 32,
+    "ṣ": 33,
+    "ṭ": 34,
+    "ẓ": 29,
+    "‐": 7
+  },
+  "shk": {
+    "'": 41,
+    "-": 27,
+    "1": 49,
+    "2": 48,
+    "8": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 37,
+    "b": 17,
+    "c": 31,
+    "d": 14,
+    "e": 10,
+    "g": 8,
+    "h": 19,
+    "i": 7,
+    "j": 20,
+    "k": 9,
+    "l": 18,
+    "m": 15,
+    "n": 6,
+    "o": 28,
+    "p": 29,
+    "r": 21,
+    "s": 34,
+    "t": 26,
+    "u": 23,
+    "w": 13,
+    "y": 11,
+    "|": 4,
+    "á": 38,
+    "ä": 46,
+    "è": 39,
+    "é": 44,
+    "ë": 16,
+    "ì": 32,
+    "í": 35,
+    "ï": 22,
+    "ó": 33,
+    "ö": 25,
+    "ø": 12,
+    "ù": 40,
+    "ú": 45,
+    "ɑ": 5,
+    "̀": 36,
+    "́": 30,
+    "̈": 24,
+    "–": 42,
+    "—": 43
+  },
+  "shn": {
+    "'": 44,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "င": 15,
+    "တ": 13,
+    "ထ": 40,
+    "ပ": 24,
+    "မ": 14,
+    "ယ": 33,
+    "ရ": 43,
+    "လ": 20,
+    "ဝ": 9,
+    "သ": 25,
+    "ဢ": 17,
+    "ိ": 19,
+    "ီ": 35,
+    "ု": 26,
+    "ူ": 16,
+    "ေ": 23,
+    "ဵ": 32,
+    "း": 8,
+    "်": 5,
+    "ျ": 42,
+    "ြ": 39,
+    "ွ": 31,
+    "ၢ": 21,
+    "ၵ": 11,
+    "ၶ": 29,
+    "ၸ": 22,
+    "ၺ": 41,
+    "ၼ": 6,
+    "ၽ": 36,
+    "ၿ": 46,
+    "ႀ": 45,
+    "ႁ": 27,
+    "ႂ": 30,
+    "ႃ": 10,
+    "ႄ": 34,
+    "ႅ": 37,
+    "ႆ": 28,
+    "ႇ": 12,
+    "ႈ": 7,
+    "ႉ": 18
+  },
+  "sho": {
+    "'": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 36,
+    "d": 13,
+    "e": 11,
+    "f": 42,
+    "g": 16,
+    "h": 21,
+    "i": 6,
+    "j": 30,
+    "k": 8,
+    "l": 20,
+    "m": 15,
+    "n": 10,
+    "o": 12,
+    "p": 34,
+    "r": 43,
+    "s": 25,
+    "t": 14,
+    "u": 19,
+    "v": 48,
+    "w": 26,
+    "y": 28,
+    "z": 24,
+    "|": 4,
+    "à": 17,
+    "á": 31,
+    "ã": 18,
+    "è": 38,
+    "é": 44,
+    "ì": 23,
+    "í": 49,
+    "ò": 37,
+    "ó": 46,
+    "ù": 40,
+    "ú": 50,
+    "ĩ": 35,
+    "ń": 45,
+    "ũ": 33,
+    "ǹ": 41,
+    "ɔ": 22,
+    "ɛ": 7,
+    "̀": 29,
+    "́": 39,
+    "̃": 27,
+    "ḿ": 47
+  },
+  "shp": {
+    "'": 35,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 13,
+    "c": 23,
+    "d": 25,
+    "e": 10,
+    "f": 33,
+    "g": 31,
+    "h": 18,
+    "i": 6,
+    "j": 11,
+    "k": 9,
+    "l": 28,
+    "m": 16,
+    "n": 7,
+    "o": 8,
+    "p": 22,
+    "q": 37,
+    "r": 15,
+    "s": 14,
+    "t": 12,
+    "u": 27,
+    "v": 36,
+    "w": 20,
+    "x": 17,
+    "y": 19,
+    "z": 34,
+    "|": 5,
+    "á": 21,
+    "é": 26,
+    "í": 24,
+    "ñ": 38,
+    "ó": 30,
+    "ú": 32,
+    "—": 29
+  },
+  "sid": {
+    "'": 22,
+    "-": 37,
+    "0": 31,
+    "1": 32,
+    "2": 33,
+    "3": 38,
+    "4": 35,
+    "5": 36,
+    "6": 34,
+    "7": 39,
+    "8": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 23,
+    "d": 18,
+    "e": 9,
+    "f": 26,
+    "g": 20,
+    "h": 11,
+    "i": 7,
+    "j": 28,
+    "k": 16,
+    "l": 17,
+    "m": 15,
+    "n": 6,
+    "o": 8,
+    "p": 29,
+    "q": 25,
+    "r": 14,
+    "s": 10,
+    "t": 13,
+    "u": 12,
+    "w": 24,
+    "x": 27,
+    "y": 19,
+    "z": 30,
+    "|": 5
+  },
+  "sig": {
+    "'": 33,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "d": 22,
+    "e": 13,
+    "f": 29,
+    "g": 23,
+    "h": 25,
+    "i": 17,
+    "k": 14,
+    "l": 11,
+    "m": 19,
+    "n": 15,
+    "o": 21,
+    "p": 28,
+    "r": 24,
+    "s": 16,
+    "t": 20,
+    "u": 26,
+    "v": 31,
+    "w": 18,
+    "y": 7,
+    "z": 34,
+    "|": 4,
+    "á": 30,
+    "ŋ": 8,
+    "ɔ": 27,
+    "ɛ": 12,
+    "ɩ": 6,
+    "ʋ": 10,
+    "̃": 35
+  },
+  "sil": {
+    "'": 35,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 29,
+    "d": 15,
+    "e": 17,
+    "f": 30,
+    "g": 28,
+    "h": 24,
+    "i": 6,
+    "j": 27,
+    "k": 18,
+    "l": 9,
+    "m": 14,
+    "n": 11,
+    "o": 21,
+    "p": 25,
+    "r": 20,
+    "s": 13,
+    "t": 16,
+    "u": 8,
+    "v": 34,
+    "w": 19,
+    "y": 26,
+    "z": 32,
+    "|": 4,
+    "á": 33,
+    "í": 31,
+    "ú": 37,
+    "ŋ": 7,
+    "ɔ": 38,
+    "ɛ": 10,
+    "́": 36,
+    "ↄ": 23
+  },
+  "sja": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 19,
+    "d": 13,
+    "e": 7,
+    "f": 43,
+    "g": 28,
+    "h": 21,
+    "i": 6,
+    "j": 18,
+    "k": 15,
+    "l": 30,
+    "m": 12,
+    "n": 22,
+    "o": 17,
+    "p": 8,
+    "q": 46,
+    "r": 9,
+    "s": 25,
+    "t": 14,
+    "u": 20,
+    "v": 41,
+    "w": 24,
+    "x": 48,
+    "y": 33,
+    "z": 45,
+    "|": 5,
+    "á": 31,
+    "ã": 11,
+    "é": 40,
+    "ë": 36,
+    "í": 42,
+    "ñ": 47,
+    "ó": 44,
+    "õ": 26,
+    "ú": 39,
+    "ĩ": 29,
+    "ũ": 34,
+    "ɨ": 16,
+    "́": 37,
+    "̃": 32,
+    "ẽ": 35,
+    "–": 27,
+    "—": 38
+  },
+  "sjm": {
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 14,
+    "c": 33,
+    "d": 20,
+    "e": 22,
+    "g": 16,
+    "h": 21,
+    "i": 7,
+    "j": 24,
+    "k": 11,
+    "l": 13,
+    "m": 9,
+    "n": 6,
+    "o": 17,
+    "p": 19,
+    "r": 25,
+    "s": 10,
+    "t": 12,
+    "u": 8,
+    "w": 23,
+    "y": 15,
+    "z": 32,
+    "|": 5,
+    "ā": 27,
+    "ē": 31,
+    "ī": 30,
+    "ō": 28,
+    "ū": 29,
+    "ꞌ": 18
+  },
+  "sld": {
+    "'": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 20,
+    "c": 36,
+    "d": 31,
+    "e": 26,
+    "f": 37,
+    "g": 40,
+    "h": 25,
+    "i": 21,
+    "j": 38,
+    "k": 19,
+    "l": 15,
+    "m": 17,
+    "n": 10,
+    "o": 29,
+    "p": 30,
+    "q": 47,
+    "r": 12,
+    "s": 14,
+    "t": 22,
+    "u": 24,
+    "v": 41,
+    "w": 13,
+    "y": 32,
+    "z": 28,
+    "|": 4,
+    "á": 9,
+    "ã": 45,
+    "é": 27,
+    "í": 23,
+    "ó": 33,
+    "õ": 46,
+    "ú": 34,
+    "ĩ": 44,
+    "ŋ": 16,
+    "ɓ": 35,
+    "ɔ": 18,
+    "ɛ": 7,
+    "ɩ": 6,
+    "ʋ": 11,
+    "́": 5,
+    "̃": 39,
+    "ṹ": 48,
+    "ẽ": 43
+  },
+  "slk": {
+    "$": 70,
+    "%": 63,
+    "&": 73,
+    "'": 64,
+    "+": 75,
+    ",": 61,
+    "-": 56,
+    ".": 47,
+    "/": 58,
+    "0": 39,
+    "1": 42,
+    "2": 45,
+    "3": 54,
+    "4": 52,
+    "5": 48,
+    "6": 53,
+    "7": 55,
+    "8": 51,
+    "9": 50,
+    ":": 59,
+    ";": 69,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "c": 20,
+    "d": 16,
+    "e": 7,
+    "f": 37,
+    "g": 36,
+    "h": 21,
+    "i": 9,
+    "j": 22,
+    "k": 14,
+    "l": 15,
+    "m": 18,
+    "n": 8,
+    "o": 5,
+    "p": 17,
+    "q": 62,
+    "r": 11,
+    "s": 12,
+    "t": 10,
+    "u": 19,
+    "v": 13,
+    "w": 49,
+    "x": 46,
+    "y": 26,
+    "z": 23,
+    "|": 4,
+    "°": 76,
+    "²": 74,
+    "á": 24,
+    "ã": 68,
+    "ä": 43,
+    "ç": 72,
+    "é": 32,
+    "í": 27,
+    "ó": 40,
+    "ô": 38,
+    "õ": 67,
+    "ú": 31,
+    "ü": 65,
+    "ý": 28,
+    "č": 30,
+    "ď": 41,
+    "ĺ": 57,
+    "ľ": 35,
+    "ň": 44,
+    "ŕ": 60,
+    "š": 33,
+    "ť": 34,
+    "ž": 29,
+    "̇": 71,
+    "–": 66
+  },
+  "slu": {
+    "'": 29,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 28,
+    "d": 18,
+    "e": 6,
+    "f": 23,
+    "g": 22,
+    "h": 19,
+    "i": 8,
+    "j": 24,
+    "k": 7,
+    "l": 17,
+    "m": 11,
+    "n": 13,
+    "o": 15,
+    "p": 26,
+    "r": 12,
+    "s": 10,
+    "t": 14,
+    "u": 16,
+    "w": 20,
+    "y": 9,
+    "z": 27,
+    "|": 4
+  },
+  "slv": {
+    "\"": 71,
+    "%": 53,
+    "&": 67,
+    "'": 52,
+    "+": 61,
+    ",": 46,
+    "-": 42,
+    ".": 41,
+    "/": 45,
+    "0": 30,
+    "1": 31,
+    "2": 32,
+    "3": 39,
+    "4": 36,
+    "5": 35,
+    "6": 40,
+    "7": 43,
+    "8": 38,
+    "9": 33,
+    ":": 51,
+    ";": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 68,
+    "]": 69,
+    "a": 5,
+    "b": 22,
+    "c": 26,
+    "d": 17,
+    "e": 6,
+    "f": 29,
+    "g": 23,
+    "h": 24,
+    "i": 8,
+    "j": 13,
+    "k": 16,
+    "l": 14,
+    "m": 19,
+    "n": 9,
+    "o": 7,
+    "p": 18,
+    "q": 47,
+    "r": 10,
+    "s": 12,
+    "t": 11,
+    "u": 21,
+    "v": 15,
+    "w": 37,
+    "x": 48,
+    "y": 34,
+    "z": 20,
+    "|": 4,
+    "«": 50,
+    "°": 64,
+    "²": 65,
+    "»": 49,
+    "á": 55,
+    "ç": 59,
+    "é": 60,
+    "í": 66,
+    "ó": 70,
+    "õ": 57,
+    "ö": 63,
+    "ú": 56,
+    "ü": 54,
+    "ć": 62,
+    "č": 25,
+    "š": 27,
+    "ž": 28,
+    "–": 44
+  },
+  "sml": {
+    "'": 10,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "d": 20,
+    "e": 22,
+    "g": 13,
+    "h": 19,
+    "i": 7,
+    "j": 27,
+    "k": 12,
+    "l": 15,
+    "m": 8,
+    "n": 6,
+    "o": 21,
+    "p": 18,
+    "r": 25,
+    "s": 11,
+    "t": 14,
+    "u": 9,
+    "w": 23,
+    "y": 17,
+    "|": 5,
+    "­": 32,
+    "ā": 24,
+    "ē": 30,
+    "ī": 31,
+    "ō": 28,
+    "ū": 29
+  },
+  "smo": {
+    "'": 14,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "e": 8,
+    "f": 15,
+    "g": 17,
+    "h": 27,
+    "i": 6,
+    "k": 23,
+    "l": 10,
+    "m": 12,
+    "n": 13,
+    "o": 7,
+    "p": 19,
+    "r": 22,
+    "s": 16,
+    "t": 11,
+    "u": 9,
+    "v": 20,
+    "|": 4,
+    "ā": 18,
+    "ē": 21,
+    "ī": 26,
+    "ō": 24,
+    "ū": 25
+  },
+  "sna": {
+    "!": 54,
+    "\"": 53,
+    "$": 47,
+    "%": 50,
+    "&": 55,
+    "'": 43,
+    ",": 42,
+    "-": 38,
+    ".": 39,
+    "/": 46,
+    "0": 29,
+    "1": 30,
+    "2": 31,
+    "3": 35,
+    "4": 34,
+    "5": 33,
+    "6": 37,
+    "7": 40,
+    "8": 36,
+    "9": 32,
+    ":": 45,
+    ";": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 48,
+    "[": 58,
+    "]": 59,
+    "_": 60,
+    "a": 4,
+    "b": 25,
+    "c": 24,
+    "d": 19,
+    "e": 8,
+    "f": 26,
+    "g": 22,
+    "h": 16,
+    "i": 6,
+    "j": 28,
+    "k": 10,
+    "l": 27,
+    "m": 13,
+    "n": 7,
+    "o": 12,
+    "p": 23,
+    "q": 44,
+    "r": 11,
+    "s": 18,
+    "t": 17,
+    "u": 9,
+    "v": 14,
+    "w": 20,
+    "x": 41,
+    "y": 21,
+    "z": 15,
+    "|": 5,
+    "°": 52,
+    "á": 57,
+    "ã": 61,
+    "ç": 56,
+    "é": 63,
+    "ú": 62,
+    "ü": 51,
+    "ʼ": 64
+  },
+  "snd": {
+    "!": 125,
+    "\"": 116,
+    "%": 108,
+    "&": 127,
+    "'": 122,
+    "+": 119,
+    ",": 99,
+    "-": 70,
+    ".": 72,
+    "/": 97,
+    "0": 46,
+    "1": 49,
+    "2": 56,
+    "3": 67,
+    "4": 65,
+    "5": 63,
+    "6": 68,
+    "7": 74,
+    "8": 69,
+    "9": 61,
+    ":": 91,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 73,
+    "b": 92,
+    "c": 75,
+    "d": 87,
+    "e": 80,
+    "f": 98,
+    "g": 90,
+    "h": 95,
+    "i": 78,
+    "j": 114,
+    "k": 104,
+    "l": 93,
+    "m": 88,
+    "n": 85,
+    "o": 79,
+    "p": 77,
+    "q": 115,
+    "r": 89,
+    "s": 76,
+    "t": 83,
+    "u": 84,
+    "v": 94,
+    "w": 106,
+    "x": 111,
+    "y": 107,
+    "z": 113,
+    "|": 4,
+    "¥": 117,
+    "°": 120,
+    "²": 123,
+    "õ": 118,
+    "،": 52,
+    "؛": 112,
+    "؟": 124,
+    "ء": 40,
+    "آ": 24,
+    "ئ": 20,
+    "ا": 6,
+    "ب": 19,
+    "ت": 15,
+    "ث": 53,
+    "ج": 12,
+    "ح": 36,
+    "خ": 39,
+    "د": 17,
+    "ذ": 58,
+    "ر": 9,
+    "ز": 33,
+    "س": 16,
+    "ش": 29,
+    "ص": 38,
+    "ض": 57,
+    "ط": 44,
+    "ظ": 54,
+    "ع": 31,
+    "غ": 60,
+    "ـ": 126,
+    "ف": 27,
+    "ق": 30,
+    "ل": 14,
+    "م": 13,
+    "ن": 7,
+    "ه": 10,
+    "و": 8,
+    "ي": 5,
+    "ً": 100,
+    "َ": 82,
+    "ُ": 62,
+    "ِ": 42,
+    "ٰ": 86,
+    "ٺ": 51,
+    "ٻ": 45,
+    "ٽ": 21,
+    "پ": 18,
+    "ٿ": 26,
+    "ڀ": 50,
+    "ڃ": 59,
+    "ڄ": 66,
+    "چ": 37,
+    "ڇ": 55,
+    "ڈ": 121,
+    "ڊ": 34,
+    "ڌ": 48,
+    "ڍ": 64,
+    "ڏ": 32,
+    "ڙ": 43,
+    "ڦ": 81,
+    "ک": 25,
+    "ڪ": 11,
+    "گ": 22,
+    "ڱ": 96,
+    "ڳ": 47,
+    "ڻ": 23,
+    "ھ": 71,
+    "ہ": 41,
+    "ی": 101,
+    "۽": 35,
+    "۾": 28,
+    "–": 102,
+    "—": 103,
+    "‘": 109,
+    "’": 110,
+    "”": 105
+  },
+  "sne": {
+    "'": 36,
+    "-": 24,
+    "0": 32,
+    "1": 35,
+    "2": 34,
+    "3": 39,
+    "4": 38,
+    "5": 42,
+    "6": 37,
+    "7": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 31,
+    "d": 10,
+    "e": 12,
+    "f": 27,
+    "g": 14,
+    "h": 13,
+    "i": 8,
+    "j": 23,
+    "k": 17,
+    "l": 25,
+    "m": 19,
+    "n": 6,
+    "o": 7,
+    "p": 20,
+    "q": 41,
+    "r": 18,
+    "s": 15,
+    "t": 16,
+    "u": 9,
+    "v": 30,
+    "w": 26,
+    "x": 33,
+    "y": 22,
+    "z": 29,
+    "|": 4,
+    "—": 28,
+    "ꞌ": 11
+  },
+  "snn": {
+    "'": 10,
+    "0": 40,
+    "1": 42,
+    "2": 43,
+    "4": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 14,
+    "d": 29,
+    "e": 9,
+    "f": 34,
+    "g": 19,
+    "h": 22,
+    "i": 6,
+    "j": 16,
+    "k": 39,
+    "l": 28,
+    "m": 21,
+    "n": 11,
+    "o": 12,
+    "p": 25,
+    "q": 24,
+    "r": 15,
+    "s": 18,
+    "t": 20,
+    "u": 8,
+    "v": 35,
+    "x": 38,
+    "y": 17,
+    "z": 26,
+    "|": 4,
+    "á": 37,
+    "é": 33,
+    "ë": 7,
+    "í": 31,
+    "ñ": 23,
+    "ó": 36,
+    "ú": 32,
+    "ü": 27,
+    "—": 30
+  },
+  "snp": {
+    "'": 25,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 17,
+    "e": 8,
+    "f": 18,
+    "g": 19,
+    "h": 26,
+    "i": 7,
+    "k": 13,
+    "l": 11,
+    "m": 10,
+    "n": 9,
+    "o": 6,
+    "p": 24,
+    "s": 20,
+    "t": 12,
+    "u": 14,
+    "w": 16,
+    "y": 15,
+    "|": 4,
+    "á": 23
+  },
+  "snw": {
+    "'": 36,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 22,
+    "e": 10,
+    "f": 21,
+    "g": 29,
+    "h": 28,
+    "i": 6,
+    "k": 8,
+    "l": 11,
+    "m": 19,
+    "n": 7,
+    "o": 9,
+    "p": 23,
+    "r": 25,
+    "s": 17,
+    "t": 16,
+    "u": 15,
+    "v": 33,
+    "w": 20,
+    "y": 18,
+    "|": 4,
+    "à": 37,
+    "á": 30,
+    "ã": 32,
+    "é": 34,
+    "í": 31,
+    "ú": 38,
+    "ĩ": 26,
+    "ũ": 24,
+    "ɔ": 13,
+    "ɛ": 12,
+    "́": 39,
+    "̃": 35
+  },
+  "som": {
+    "!": 62,
+    "\"": 51,
+    "$": 49,
+    "%": 53,
+    "&": 54,
+    "'": 35,
+    "+": 57,
+    ",": 42,
+    "-": 32,
+    ".": 39,
+    "/": 46,
+    "0": 28,
+    "1": 29,
+    "2": 31,
+    "3": 38,
+    "4": 37,
+    "5": 36,
+    "6": 40,
+    "7": 44,
+    "8": 41,
+    "9": 34,
+    ":": 45,
+    ";": 52,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 66,
+    "]": 67,
+    "_": 64,
+    "a": 4,
+    "b": 18,
+    "c": 24,
+    "d": 8,
+    "e": 11,
+    "f": 26,
+    "g": 19,
+    "h": 14,
+    "i": 6,
+    "j": 27,
+    "k": 15,
+    "l": 17,
+    "m": 20,
+    "n": 9,
+    "o": 7,
+    "p": 30,
+    "q": 25,
+    "r": 16,
+    "s": 13,
+    "t": 21,
+    "u": 12,
+    "v": 33,
+    "w": 22,
+    "x": 23,
+    "y": 10,
+    "z": 43,
+    "|": 5,
+    "£": 63,
+    "¥": 55,
+    "°": 59,
+    "²": 69,
+    "á": 60,
+    "ç": 68,
+    "é": 72,
+    "í": 71,
+    "ö": 70,
+    "ú": 58,
+    "ü": 50,
+    "​": 61,
+    "—": 65,
+    "‘": 56,
+    "’": 47,
+    "”": 48
+  },
+  "soy": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 33,
+    "c": 22,
+    "e": 19,
+    "f": 28,
+    "h": 29,
+    "i": 6,
+    "k": 10,
+    "l": 17,
+    "m": 15,
+    "n": 8,
+    "o": 27,
+    "p": 11,
+    "r": 18,
+    "s": 16,
+    "t": 20,
+    "u": 12,
+    "w": 25,
+    "y": 14,
+    "|": 4,
+    "á": 37,
+    "é": 23,
+    "ë": 46,
+    "í": 21,
+    "ñ": 45,
+    "ó": 30,
+    "õ": 40,
+    "ø": 48,
+    "ú": 26,
+    "û": 44,
+    "ĩ": 39,
+    "ń": 34,
+    "ŋ": 31,
+    "ũ": 38,
+    "ɑ": 7,
+    "ɔ": 13,
+    "ɛ": 5,
+    "̀": 42,
+    "́": 9,
+    "̃": 24,
+    "ḿ": 35,
+    "ṍ": 41,
+    "ṹ": 43,
+    "ẽ": 36,
+    "–": 47
+  },
+  "spa": {
+    "!": 60,
+    "$": 62,
+    "%": 53,
+    "&": 70,
+    "'": 61,
+    "+": 65,
+    ",": 58,
+    "-": 51,
+    ".": 47,
+    "/": 55,
+    "0": 33,
+    "1": 35,
+    "2": 38,
+    "3": 45,
+    "4": 44,
+    "5": 43,
+    "6": 46,
+    "7": 49,
+    "8": 48,
+    "9": 41,
+    ":": 50,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 57,
+    "[": 67,
+    "]": 68,
+    "`": 106,
+    "a": 6,
+    "b": 20,
+    "c": 15,
+    "d": 13,
+    "e": 5,
+    "f": 22,
+    "g": 19,
+    "h": 24,
+    "i": 11,
+    "j": 29,
+    "k": 36,
+    "l": 12,
+    "m": 17,
+    "n": 9,
+    "o": 7,
+    "p": 18,
+    "q": 23,
+    "r": 10,
+    "s": 8,
+    "t": 14,
+    "u": 16,
+    "v": 21,
+    "w": 42,
+    "x": 32,
+    "y": 25,
+    "z": 30,
+    "|": 4,
+    "¡": 59,
+    "«": 40,
+    "­": 242,
+    "°": 63,
+    "º": 71,
+    "»": 39,
+    "¿": 56,
+    "ß": 92,
+    "à": 132,
+    "á": 28,
+    "â": 81,
+    "ã": 74,
+    "ä": 76,
+    "å": 90,
+    "æ": 97,
+    "ç": 66,
+    "é": 31,
+    "ê": 85,
+    "ë": 77,
+    "ì": 123,
+    "í": 27,
+    "î": 110,
+    "ï": 95,
+    "ð": 93,
+    "ñ": 34,
+    "ò": 96,
+    "ó": 26,
+    "ô": 86,
+    "õ": 64,
+    "ö": 69,
+    "ø": 82,
+    "ù": 148,
+    "ú": 37,
+    "û": 133,
+    "ü": 52,
+    "ý": 103,
+    "þ": 115,
+    "ā": 84,
+    "ă": 117,
+    "ć": 87,
+    "č": 91,
+    "đ": 114,
+    "ē": 201,
+    "ė": 111,
+    "ę": 119,
+    "ě": 130,
+    "ğ": 232,
+    "ī": 107,
+    "ı": 155,
+    "ł": 72,
+    "ń": 109,
+    "ō": 73,
+    "ŏ": 159,
+    "ő": 118,
+    "œ": 113,
+    "ř": 122,
+    "ś": 102,
+    "ş": 89,
+    "š": 83,
+    "ū": 80,
+    "ź": 121,
+    "ž": 105,
+    "ș": 134,
+    "ț": 157,
+    "ə": 88,
+    "ʻ": 94,
+    "ʽ": 211,
+    "ʿ": 139,
+    "ː": 101,
+    "́": 79,
+    "̇": 146,
+    "̈": 112,
+    "̪": 200,
+    "ϙ": 183,
+    "а": 108,
+    "б": 125,
+    "в": 124,
+    "г": 149,
+    "д": 153,
+    "е": 100,
+    "и": 99,
+    "й": 186,
+    "к": 128,
+    "л": 98,
+    "н": 127,
+    "о": 129,
+    "п": 162,
+    "р": 104,
+    "с": 150,
+    "т": 126,
+    "ч": 152,
+    "ш": 138,
+    "ы": 163,
+    "ь": 151,
+    "ю": 233,
+    "я": 234,
+    "ё": 145,
+    "ү": 137,
+    "ө": 161,
+    "ְ": 120,
+    "ִ": 169,
+    "ֵ": 172,
+    "ָ": 174,
+    "ֹ": 175,
+    "ּ": 173,
+    "ב": 213,
+    "ה": 220,
+    "ו": 131,
+    "י": 219,
+    "כ": 140,
+    "ל": 142,
+    "ם": 176,
+    "מ": 171,
+    "נ": 141,
+    "ס": 170,
+    "ק": 215,
+    "ר": 214,
+    "ש": 143,
+    "ת": 168,
+    "ب": 204,
+    "ة": 206,
+    "ذ": 202,
+    "ه": 203,
+    "ي": 205,
+    "ਆ": 224,
+    "ਘ": 227,
+    "ਤ": 229,
+    "ਨ": 230,
+    "ਮ": 228,
+    "ਸ": 135,
+    "ਾ": 136,
+    "ਿ": 225,
+    "ੰ": 226,
+    "ḫ": 144,
+    "ṁ": 210,
+    "ṃ": 185,
+    "ṇ": 216,
+    "‐": 156,
+    "–": 78,
+    "—": 75,
+    "′": 116,
+    "⃗": 164,
+    "♭": 147,
+    "し": 193,
+    "の": 184,
+    "ら": 187,
+    "ゴ": 154,
+    "ミ": 190,
+    "ム": 194,
+    "ラ": 188,
+    "㓁": 238,
+    "下": 179,
+    "仙": 182,
+    "口": 197,
+    "周": 231,
+    "妻": 180,
+    "寛": 207,
+    "山": 196,
+    "市": 181,
+    "戌": 195,
+    "日": 217,
+    "本": 218,
+    "毵": 160,
+    "浩": 209,
+    "消": 192,
+    "生": 199,
+    "申": 212,
+    "真": 198,
+    "箱": 191,
+    "网": 235,
+    "罒": 236,
+    "罓": 237,
+    "肋": 221,
+    "肌": 222,
+    "背": 223,
+    "良": 189,
+    "裕": 208,
+    "貧": 239,
+    "貨": 240,
+    "販": 241,
+    "飢": 165,
+    "飧": 166,
+    "飲": 167,
+    "鮓": 178,
+    "鮨": 177,
+    "": 158
+  },
+  "spp": {
+    "'": 31,
+    "-": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 28,
+    "c": 29,
+    "d": 38,
+    "e": 9,
+    "f": 30,
+    "g": 25,
+    "h": 22,
+    "i": 5,
+    "j": 32,
+    "k": 11,
+    "l": 18,
+    "m": 12,
+    "n": 7,
+    "o": 23,
+    "p": 13,
+    "r": 14,
+    "s": 16,
+    "t": 24,
+    "u": 10,
+    "v": 50,
+    "w": 17,
+    "y": 8,
+    "z": 41,
+    "|": 4,
+    "à": 19,
+    "á": 26,
+    "â": 48,
+    "è": 36,
+    "é": 42,
+    "ê": 47,
+    "ì": 35,
+    "í": 34,
+    "î": 44,
+    "ò": 43,
+    "ó": 51,
+    "ô": 52,
+    "ù": 33,
+    "ú": 39,
+    "û": 46,
+    "ŋ": 20,
+    "ɔ": 27,
+    "ɛ": 15,
+    "ɲ": 21,
+    "̀": 37,
+    "́": 45,
+    "̂": 49,
+    "‐": 40
+  },
+  "spy": {
+    "'": 25,
+    "-": 27,
+    "0": 32,
+    "1": 29,
+    "2": 30,
+    "6": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 11,
+    "b": 18,
+    "c": 19,
+    "e": 14,
+    "f": 28,
+    "g": 26,
+    "h": 20,
+    "i": 8,
+    "k": 5,
+    "l": 23,
+    "m": 15,
+    "n": 12,
+    "o": 6,
+    "r": 21,
+    "s": 22,
+    "t": 13,
+    "u": 10,
+    "w": 24,
+    "y": 9,
+    "|": 4,
+    "ā": 16,
+    "ē": 7,
+    "ō": 17
+  },
+  "sqi": {
+    "'": 30,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 23,
+    "c": 31,
+    "d": 16,
+    "e": 6,
+    "f": 27,
+    "g": 22,
+    "h": 13,
+    "i": 8,
+    "j": 14,
+    "k": 20,
+    "l": 21,
+    "m": 18,
+    "n": 11,
+    "o": 17,
+    "p": 19,
+    "q": 24,
+    "r": 9,
+    "s": 12,
+    "t": 7,
+    "u": 15,
+    "v": 25,
+    "x": 32,
+    "y": 28,
+    "z": 26,
+    "|": 4,
+    "à": 36,
+    "ç": 29,
+    "è": 34,
+    "ë": 5,
+    "–": 35
+  },
+  "sri": {
+    "'": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 36,
+    "d": 21,
+    "e": 8,
+    "f": 41,
+    "g": 13,
+    "h": 40,
+    "i": 9,
+    "j": 24,
+    "k": 18,
+    "l": 37,
+    "m": 11,
+    "n": 28,
+    "o": 15,
+    "p": 17,
+    "q": 48,
+    "r": 5,
+    "s": 14,
+    "t": 26,
+    "u": 16,
+    "v": 42,
+    "w": 27,
+    "x": 49,
+    "y": 22,
+    "z": 47,
+    "|": 4,
+    "á": 33,
+    "ã": 7,
+    "é": 34,
+    "í": 31,
+    "ñ": 30,
+    "ó": 39,
+    "õ": 25,
+    "ú": 35,
+    "ü": 43,
+    "ĩ": 12,
+    "ũ": 29,
+    "ʉ": 10,
+    "́": 32,
+    "̃": 19,
+    "̶": 46,
+    "ṍ": 45,
+    "ẽ": 23,
+    "—": 38
+  },
+  "srm": {
+    "'": 29,
+    "-": 30,
+    "0": 32,
+    "1": 34,
+    "2": 35,
+    "4": 36,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 9,
+    "e": 10,
+    "f": 19,
+    "g": 20,
+    "h": 22,
+    "i": 6,
+    "j": 24,
+    "k": 15,
+    "l": 23,
+    "m": 17,
+    "n": 7,
+    "o": 16,
+    "p": 25,
+    "r": 39,
+    "s": 12,
+    "t": 14,
+    "u": 8,
+    "v": 31,
+    "w": 21,
+    "z": 28,
+    "|": 4,
+    "ë": 11,
+    "í": 37,
+    "ö": 13,
+    "ú": 33,
+    "˻": 26,
+    "˼": 27
+  },
+  "srn": {
+    "'": 24,
+    "-": 27,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 37,
+    "d": 11,
+    "e": 8,
+    "f": 17,
+    "g": 19,
+    "h": 26,
+    "i": 7,
+    "j": 32,
+    "k": 14,
+    "l": 22,
+    "m": 16,
+    "n": 6,
+    "o": 12,
+    "p": 21,
+    "r": 13,
+    "s": 10,
+    "t": 15,
+    "u": 9,
+    "v": 29,
+    "w": 20,
+    "y": 23,
+    "z": 35,
+    "|": 4,
+    "á": 36,
+    "è": 25,
+    "ê": 28,
+    "ô": 30
+  },
+  "srp-script_cyrillic": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 22,
+    "в": 16,
+    "г": 24,
+    "д": 14,
+    "е": 7,
+    "ж": 25,
+    "з": 23,
+    "и": 8,
+    "к": 18,
+    "л": 19,
+    "м": 11,
+    "н": 9,
+    "о": 6,
+    "п": 20,
+    "р": 15,
+    "с": 12,
+    "т": 10,
+    "у": 17,
+    "ф": 33,
+    "х": 26,
+    "ц": 27,
+    "ч": 30,
+    "ш": 21,
+    "ђ": 32,
+    "ј": 13,
+    "љ": 28,
+    "њ": 29,
+    "ћ": 31,
+    "џ": 34
+  },
+  "srp-script_latin": {
+    "!": 89,
+    "%": 79,
+    "'": 86,
+    "+": 90,
+    ",": 75,
+    "-": 63,
+    ".": 53,
+    "/": 73,
+    "0": 48,
+    "1": 50,
+    "2": 54,
+    "3": 67,
+    "4": 65,
+    "5": 62,
+    "6": 66,
+    "7": 71,
+    "8": 68,
+    "9": 59,
+    ":": 74,
+    ";": 82,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 87,
+    "]": 88,
+    "a": 5,
+    "b": 23,
+    "c": 30,
+    "d": 16,
+    "e": 8,
+    "f": 46,
+    "g": 22,
+    "h": 32,
+    "i": 6,
+    "j": 14,
+    "k": 15,
+    "l": 18,
+    "m": 19,
+    "n": 9,
+    "o": 7,
+    "p": 20,
+    "q": 92,
+    "r": 10,
+    "s": 11,
+    "t": 12,
+    "u": 13,
+    "v": 17,
+    "w": 78,
+    "x": 80,
+    "y": 84,
+    "z": 21,
+    "|": 4,
+    "°": 83,
+    "²": 91,
+    "×": 95,
+    "õ": 85,
+    "ć": 36,
+    "č": 28,
+    "đ": 47,
+    "š": 29,
+    "ž": 33,
+    "а": 24,
+    "б": 52,
+    "в": 41,
+    "г": 51,
+    "д": 39,
+    "е": 25,
+    "ж": 61,
+    "з": 49,
+    "и": 26,
+    "к": 40,
+    "л": 42,
+    "м": 44,
+    "н": 31,
+    "о": 27,
+    "п": 45,
+    "р": 34,
+    "с": 38,
+    "т": 35,
+    "у": 37,
+    "ф": 70,
+    "х": 64,
+    "ц": 57,
+    "ч": 55,
+    "ш": 56,
+    "ђ": 72,
+    "ј": 43,
+    "љ": 60,
+    "њ": 58,
+    "ћ": 69,
+    "џ": 76,
+    "​": 81,
+    "–": 94,
+    "—": 93,
+    "”": 77
+  },
+  "srx": {
+    "'": 62,
+    "-": 59,
+    "0": 64,
+    "1": 66,
+    "4": 65,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 58,
+    "ं": 19,
+    "अ": 48,
+    "आ": 33,
+    "इ": 45,
+    "ई": 41,
+    "उ": 60,
+    "ऊ": 52,
+    "ए": 32,
+    "ऐ": 39,
+    "ओ": 50,
+    "औ": 28,
+    "क": 9,
+    "ख": 34,
+    "ग": 37,
+    "घ": 56,
+    "च": 40,
+    "छ": 47,
+    "ज": 17,
+    "झ": 57,
+    "ञ": 61,
+    "ट": 43,
+    "ठ": 51,
+    "ड": 42,
+    "ढ": 54,
+    "ण": 27,
+    "त": 10,
+    "थ": 35,
+    "द": 13,
+    "ध": 46,
+    "न": 23,
+    "प": 22,
+    "फ": 49,
+    "ब": 18,
+    "भ": 44,
+    "म": 16,
+    "य": 25,
+    "र": 8,
+    "ल": 26,
+    "व": 24,
+    "श": 21,
+    "ष": 55,
+    "स": 12,
+    "ह": 30,
+    "़": 38,
+    "ा": 6,
+    "ि": 14,
+    "ी": 11,
+    "ु": 15,
+    "ू": 36,
+    "ृ": 63,
+    "े": 29,
+    "ै": 5,
+    "ो": 20,
+    "ौ": 7,
+    "्": 31,
+    "‍": 53
+  },
+  "stn": {
+    "'": 27,
+    "3": 29,
+    "5": 28,
+    "9": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 25,
+    "d": 22,
+    "e": 14,
+    "f": 16,
+    "g": 11,
+    "h": 21,
+    "i": 6,
+    "j": 23,
+    "k": 10,
+    "l": 24,
+    "m": 12,
+    "n": 7,
+    "o": 9,
+    "p": 19,
+    "q": 20,
+    "r": 8,
+    "s": 18,
+    "t": 15,
+    "u": 13,
+    "v": 26,
+    "w": 17,
+    "|": 5
+  },
+  "stp": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 22,
+    "d": 18,
+    "e": 27,
+    "f": 30,
+    "g": 14,
+    "h": 11,
+    "i": 7,
+    "j": 10,
+    "k": 15,
+    "l": 26,
+    "m": 9,
+    "n": 12,
+    "o": 20,
+    "p": 19,
+    "r": 24,
+    "s": 25,
+    "t": 16,
+    "u": 8,
+    "x": 21,
+    "y": 29,
+    "|": 4,
+    "ñ": 23,
+    "ɇ": 28,
+    "ɨ": 13
+  },
+  "suc": {
+    "'": 25,
+    "0": 27,
+    "5": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 19,
+    "d": 16,
+    "e": 23,
+    "g": 8,
+    "h": 24,
+    "i": 9,
+    "k": 12,
+    "l": 15,
+    "m": 11,
+    "n": 6,
+    "o": 5,
+    "p": 18,
+    "r": 22,
+    "s": 14,
+    "t": 13,
+    "u": 10,
+    "w": 21,
+    "y": 20,
+    "|": 4,
+    "—": 26,
+    "ꞌ": 17
+  },
+  "suk": {
+    "'": 24,
+    "-": 31,
+    "0": 36,
+    "1": 37,
+    "2": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "c": 29,
+    "d": 25,
+    "e": 11,
+    "f": 28,
+    "g": 10,
+    "h": 17,
+    "i": 12,
+    "j": 23,
+    "k": 15,
+    "l": 8,
+    "m": 18,
+    "n": 6,
+    "o": 13,
+    "p": 27,
+    "s": 21,
+    "t": 22,
+    "u": 20,
+    "v": 32,
+    "w": 16,
+    "y": 19,
+    "z": 26,
+    "|": 4,
+    "à": 34,
+    "è": 35,
+    "ì": 39,
+    "ò": 30,
+    "ó": 33,
+    "ī": 9,
+    "ū": 7
+  },
+  "sun": {
+    "'": 28,
+    "-": 25,
+    "0": 31,
+    "1": 33,
+    "2": 30,
+    "3": 38,
+    "4": 36,
+    "5": 35,
+    "6": 34,
+    "7": 39,
+    "8": 37,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 18,
+    "e": 8,
+    "f": 29,
+    "g": 11,
+    "h": 16,
+    "i": 9,
+    "j": 21,
+    "k": 10,
+    "l": 15,
+    "m": 17,
+    "n": 6,
+    "o": 22,
+    "p": 19,
+    "r": 12,
+    "s": 14,
+    "t": 13,
+    "u": 7,
+    "w": 24,
+    "y": 23,
+    "z": 32,
+    "|": 5,
+    "—": 27
+  },
+  "sur": {
+    "'": 39,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 33,
+    "a": 6,
+    "b": 25,
+    "c": 31,
+    "d": 27,
+    "e": 8,
+    "f": 28,
+    "g": 17,
+    "h": 21,
+    "i": 5,
+    "j": 26,
+    "k": 11,
+    "l": 22,
+    "m": 9,
+    "n": 7,
+    "o": 12,
+    "p": 20,
+    "r": 14,
+    "s": 13,
+    "t": 19,
+    "u": 10,
+    "v": 38,
+    "w": 16,
+    "y": 23,
+    "z": 32,
+    "|": 4,
+    "à": 37,
+    "á": 35,
+    "é": 41,
+    "ì": 30,
+    "í": 34,
+    "ú": 40,
+    "ɓ": 24,
+    "ɗ": 15,
+    "̵": 29,
+    "̶": 18
+  },
+  "sus": {
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 23,
+    "e": 9,
+    "f": 21,
+    "g": 25,
+    "h": 27,
+    "i": 7,
+    "k": 22,
+    "l": 16,
+    "m": 10,
+    "n": 6,
+    "o": 15,
+    "p": 28,
+    "r": 12,
+    "s": 18,
+    "t": 20,
+    "u": 13,
+    "w": 24,
+    "x": 8,
+    "y": 19,
+    "|": 4,
+    "ɔ": 17,
+    "ɛ": 11,
+    "ɲ": 26
+  },
+  "suv": {
+    "'": 29,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 15,
+    "c": 24,
+    "d": 21,
+    "e": 8,
+    "f": 28,
+    "g": 14,
+    "h": 20,
+    "i": 6,
+    "j": 17,
+    "k": 13,
+    "l": 16,
+    "m": 23,
+    "n": 7,
+    "o": 9,
+    "p": 22,
+    "r": 11,
+    "s": 19,
+    "t": 12,
+    "u": 10,
+    "v": 30,
+    "w": 26,
+    "y": 25,
+    "z": 31,
+    "|": 5,
+    "ë": 18,
+    "ü": 27,
+    "ǃ": 32,
+    "ː": 33,
+    "̈": 34
+  },
+  "suz": {
+    "'": 69,
+    "-": 64,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ं": 22,
+    "अ": 41,
+    "आ": 26,
+    "इ": 21,
+    "ई": 67,
+    "उ": 49,
+    "ऊ": 57,
+    "ए": 43,
+    "ऐ": 66,
+    "ओ": 46,
+    "औ": 68,
+    "क": 8,
+    "ख": 34,
+    "ग": 28,
+    "घ": 60,
+    "ङ": 31,
+    "च": 30,
+    "छ": 44,
+    "ज": 37,
+    "झ": 62,
+    "ट": 42,
+    "ठ": 39,
+    "ड": 45,
+    "ढ": 48,
+    "त": 20,
+    "थ": 33,
+    "द": 25,
+    "ध": 50,
+    "न": 12,
+    "प": 16,
+    "फ": 38,
+    "ब": 19,
+    "भ": 36,
+    "म": 7,
+    "य": 24,
+    "र": 15,
+    "ल": 18,
+    "व": 29,
+    "श": 17,
+    "स": 23,
+    "ह": 32,
+    "़": 35,
+    "ा": 5,
+    "ि": 27,
+    "ी": 11,
+    "ु": 13,
+    "ू": 40,
+    "े": 10,
+    "ै": 51,
+    "ो": 14,
+    "ौ": 47,
+    "्": 6,
+    "०": 54,
+    "१": 52,
+    "२": 56,
+    "३": 61,
+    "४": 55,
+    "५": 58,
+    "६": 59,
+    "७": 53,
+    "८": 63,
+    "९": 65,
+    "‍": 9
+  },
+  "swe": {
+    "!": 56,
+    "\"": 30,
+    "$": 70,
+    "%": 55,
+    "&": 60,
+    "'": 51,
+    "+": 58,
+    ",": 47,
+    "-": 33,
+    ".": 34,
+    "/": 50,
+    "0": 31,
+    "1": 32,
+    "2": 36,
+    "3": 41,
+    "4": 40,
+    "5": 39,
+    "6": 43,
+    "7": 45,
+    "8": 44,
+    "9": 38,
+    ":": 46,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 72,
+    "[": 67,
+    "]": 68,
+    "a": 6,
+    "b": 26,
+    "c": 27,
+    "d": 13,
+    "e": 5,
+    "f": 19,
+    "g": 17,
+    "h": 23,
+    "i": 11,
+    "j": 29,
+    "k": 16,
+    "l": 12,
+    "m": 15,
+    "n": 9,
+    "o": 14,
+    "p": 21,
+    "q": 48,
+    "r": 7,
+    "s": 10,
+    "t": 8,
+    "u": 22,
+    "v": 18,
+    "w": 37,
+    "x": 35,
+    "y": 28,
+    "z": 42,
+    "|": 4,
+    "°": 63,
+    "²": 71,
+    "á": 57,
+    "ã": 73,
+    "ä": 20,
+    "å": 24,
+    "ç": 64,
+    "é": 49,
+    "í": 54,
+    "ó": 65,
+    "ô": 69,
+    "õ": 59,
+    "ö": 25,
+    "ú": 62,
+    "̇": 66,
+    "–": 61,
+    "”": 52
+  },
+  "swh": {
+    "!": 52,
+    "\"": 45,
+    "%": 59,
+    "&": 65,
+    "'": 48,
+    "+": 61,
+    ",": 41,
+    "-": 39,
+    ".": 31,
+    "/": 49,
+    "0": 29,
+    "1": 30,
+    "2": 32,
+    "3": 36,
+    "4": 35,
+    "5": 34,
+    "6": 37,
+    "7": 40,
+    "8": 38,
+    "9": 33,
+    ":": 43,
+    ";": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 51,
+    "[": 63,
+    "]": 64,
+    "a": 4,
+    "b": 21,
+    "c": 27,
+    "d": 22,
+    "e": 11,
+    "f": 26,
+    "g": 23,
+    "h": 14,
+    "i": 6,
+    "j": 25,
+    "k": 9,
+    "l": 16,
+    "m": 10,
+    "n": 8,
+    "o": 13,
+    "p": 24,
+    "q": 50,
+    "r": 19,
+    "s": 17,
+    "t": 15,
+    "u": 7,
+    "v": 28,
+    "w": 12,
+    "x": 42,
+    "y": 18,
+    "z": 20,
+    "|": 5,
+    "°": 60,
+    "º": 66,
+    "á": 55,
+    "â": 70,
+    "ã": 74,
+    "å": 75,
+    "ç": 62,
+    "é": 69,
+    "í": 68,
+    "ï": 71,
+    "ó": 77,
+    "õ": 57,
+    "ø": 72,
+    "ú": 67,
+    "ü": 56,
+    "ū": 76,
+    "ʼ": 78,
+    "μ": 73,
+    "–": 54,
+    "—": 58,
+    "‘": 53,
+    "’": 47,
+    "”": 44
+  },
+  "sxb": {
+    "'": 25,
+    "-": 24,
+    "0": 31,
+    "1": 32,
+    "2": 34,
+    "4": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 30,
+    "d": 21,
+    "e": 11,
+    "f": 23,
+    "g": 14,
+    "h": 29,
+    "i": 6,
+    "j": 27,
+    "k": 12,
+    "l": 17,
+    "m": 13,
+    "n": 8,
+    "o": 7,
+    "p": 26,
+    "r": 15,
+    "s": 19,
+    "t": 16,
+    "u": 9,
+    "v": 28,
+    "w": 10,
+    "y": 18,
+    "z": 20,
+    "|": 5
+  },
+  "sxn": {
+    "'": 31,
+    "-": 33,
+    "0": 38,
+    "1": 41,
+    "2": 43,
+    "4": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 37,
+    "d": 26,
+    "e": 7,
+    "f": 35,
+    "g": 10,
+    "h": 23,
+    "i": 9,
+    "j": 34,
+    "k": 11,
+    "l": 16,
+    "m": 13,
+    "n": 6,
+    "o": 19,
+    "p": 21,
+    "r": 15,
+    "s": 12,
+    "t": 14,
+    "u": 8,
+    "w": 22,
+    "y": 32,
+    "z": 36,
+    "|": 4,
+    "ó": 42,
+    "ú": 39,
+    "ě": 18,
+    "̌": 20,
+    "᷊": 27,
+    "ạ": 24,
+    "ẹ": 17,
+    "ị": 30,
+    "ọ": 29,
+    "ụ": 28
+  },
+  "sya": {
+    "'": 27,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 22,
+    "d": 21,
+    "e": 15,
+    "g": 19,
+    "h": 10,
+    "i": 8,
+    "j": 25,
+    "k": 11,
+    "l": 18,
+    "m": 13,
+    "n": 7,
+    "o": 6,
+    "p": 17,
+    "r": 14,
+    "s": 16,
+    "t": 12,
+    "u": 9,
+    "w": 24,
+    "y": 20,
+    "|": 4,
+    "–": 28
+  },
+  "syl": {
+    "'": 33,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 27,
+    "d": 19,
+    "e": 9,
+    "f": 23,
+    "g": 22,
+    "h": 18,
+    "i": 7,
+    "j": 29,
+    "k": 21,
+    "l": 13,
+    "m": 14,
+    "n": 10,
+    "o": 6,
+    "p": 30,
+    "r": 8,
+    "s": 15,
+    "t": 11,
+    "u": 12,
+    "y": 31,
+    "z": 20,
+    "|": 4,
+    "ʼ": 32,
+    "ḍ": 28,
+    "ḳ": 16,
+    "ṛ": 26,
+    "ṭ": 24,
+    "—": 34
+  },
+  "sza": {
+    "'": 27,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 23,
+    "d": 17,
+    "e": 6,
+    "g": 20,
+    "h": 11,
+    "i": 16,
+    "j": 24,
+    "k": 7,
+    "l": 15,
+    "m": 8,
+    "n": 9,
+    "o": 10,
+    "p": 21,
+    "r": 14,
+    "s": 12,
+    "t": 13,
+    "u": 18,
+    "w": 25,
+    "y": 19,
+    "|": 4,
+    "–": 28
+  },
+  "tac": {
+    "'": 13,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 20,
+    "d": 33,
+    "e": 8,
+    "f": 31,
+    "g": 14,
+    "h": 21,
+    "i": 6,
+    "j": 27,
+    "k": 18,
+    "l": 7,
+    "m": 12,
+    "n": 11,
+    "o": 16,
+    "p": 25,
+    "r": 10,
+    "s": 19,
+    "t": 26,
+    "u": 9,
+    "v": 34,
+    "w": 24,
+    "y": 29,
+    "|": 4,
+    "á": 17,
+    "é": 15,
+    "í": 23,
+    "ñ": 32,
+    "ó": 30,
+    "ú": 28,
+    "—": 35
+  },
+  "taj": {
+    "'": 59,
+    "-": 63,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "v": 64,
+    "|": 4,
+    "ँ": 54,
+    "ं": 57,
+    "ः": 43,
+    "अ": 45,
+    "आ": 36,
+    "इ": 35,
+    "ई": 60,
+    "उ": 46,
+    "ऊ": 56,
+    "ए": 37,
+    "ऐ": 61,
+    "ओ": 41,
+    "औ": 65,
+    "क": 32,
+    "ख": 33,
+    "ग": 19,
+    "घ": 48,
+    "ङ": 20,
+    "च": 28,
+    "छ": 34,
+    "ज": 26,
+    "झ": 49,
+    "ञ": 55,
+    "ट": 42,
+    "ठ": 50,
+    "ड": 51,
+    "ढ": 52,
+    "ण": 53,
+    "त": 24,
+    "थ": 27,
+    "द": 16,
+    "ध": 39,
+    "न": 9,
+    "प": 25,
+    "फ": 44,
+    "ब": 17,
+    "भ": 29,
+    "म": 10,
+    "य": 22,
+    "र": 11,
+    "ल": 12,
+    "व": 31,
+    "श": 30,
+    "ष": 47,
+    "स": 13,
+    "ह": 18,
+    "ा": 5,
+    "ि": 15,
+    "ी": 14,
+    "ु": 21,
+    "ू": 38,
+    "े": 7,
+    "ै": 40,
+    "ो": 23,
+    "ौ": 58,
+    "्": 6,
+    "६": 62,
+    "‍": 8
+  },
+  "tam": {
+    "!": 95,
+    "\"": 41,
+    "$": 89,
+    "%": 94,
+    "&": 103,
+    "'": 81,
+    "+": 97,
+    ",": 63,
+    "-": 47,
+    ".": 44,
+    "/": 78,
+    "0": 40,
+    "1": 42,
+    "2": 46,
+    "3": 55,
+    "4": 53,
+    "5": 52,
+    "6": 56,
+    "7": 60,
+    "8": 57,
+    "9": 51,
+    ":": 76,
+    ";": 91,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 102,
+    "[": 108,
+    "]": 109,
+    "`": 112,
+    "a": 61,
+    "b": 82,
+    "c": 72,
+    "d": 87,
+    "e": 65,
+    "f": 85,
+    "g": 84,
+    "h": 77,
+    "i": 70,
+    "j": 96,
+    "k": 80,
+    "l": 79,
+    "m": 64,
+    "n": 71,
+    "o": 62,
+    "p": 75,
+    "q": 98,
+    "r": 69,
+    "s": 66,
+    "t": 68,
+    "u": 74,
+    "v": 88,
+    "w": 86,
+    "x": 90,
+    "y": 92,
+    "z": 100,
+    "|": 5,
+    "°": 101,
+    "²": 105,
+    "á": 99,
+    "é": 118,
+    "ô": 117,
+    "õ": 106,
+    "š": 114,
+    "ஃ": 50,
+    "அ": 24,
+    "ஆ": 37,
+    "இ": 28,
+    "ஈ": 67,
+    "உ": 34,
+    "ஊ": 73,
+    "எ": 32,
+    "ஏ": 45,
+    "ஐ": 59,
+    "ஒ": 38,
+    "ஓ": 58,
+    "ஔ": 113,
+    "க": 6,
+    "ங": 30,
+    "ச": 22,
+    "ஜ": 43,
+    "ஞ": 54,
+    "ட": 12,
+    "ண": 26,
+    "த": 9,
+    "ந": 23,
+    "ன": 17,
+    "ப": 10,
+    "ம": 11,
+    "ய": 21,
+    "ர": 13,
+    "ற": 18,
+    "ல": 15,
+    "ள": 19,
+    "ழ": 33,
+    "வ": 16,
+    "ஷ": 49,
+    "ஸ": 39,
+    "ஹ": 48,
+    "ா": 14,
+    "ி": 8,
+    "ீ": 35,
+    "ு": 7,
+    "ூ": 36,
+    "ெ": 25,
+    "ே": 27,
+    "ை": 20,
+    "ொ": 31,
+    "ோ": 29,
+    "ௌ": 83,
+    "்": 4,
+    "ௗ": 116,
+    "ഥ": 119,
+    "​": 93,
+    "–": 111,
+    "—": 107,
+    "’": 104,
+    "”": 110,
+    "◯": 115
+  },
+  "tao": {
+    "'": 28,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 26,
+    "c": 22,
+    "d": 16,
+    "e": 13,
+    "g": 17,
+    "h": 25,
+    "i": 8,
+    "j": 21,
+    "k": 11,
+    "l": 20,
+    "m": 9,
+    "n": 6,
+    "o": 7,
+    "p": 18,
+    "r": 15,
+    "s": 14,
+    "t": 12,
+    "v": 24,
+    "w": 19,
+    "y": 10,
+    "z": 23,
+    "|": 4
+  },
+  "tap": {
+    "'": 27,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "3": 33,
+    "4": 29,
+    "5": 32,
+    "6": 35,
+    "7": 36,
+    "8": 37,
+    "9": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 38,
+    "d": 22,
+    "e": 8,
+    "f": 24,
+    "g": 21,
+    "h": 26,
+    "i": 6,
+    "k": 10,
+    "l": 11,
+    "m": 13,
+    "n": 9,
+    "o": 14,
+    "p": 18,
+    "r": 25,
+    "s": 17,
+    "t": 16,
+    "u": 7,
+    "v": 23,
+    "w": 15,
+    "y": 19,
+    "z": 20,
+    "|": 4
+  },
+  "taq": {
+    "'": 33,
+    "-": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "c": 40,
+    "d": 11,
+    "e": 16,
+    "f": 27,
+    "g": 32,
+    "h": 21,
+    "i": 9,
+    "j": 22,
+    "k": 20,
+    "l": 14,
+    "m": 13,
+    "n": 6,
+    "o": 28,
+    "q": 31,
+    "r": 19,
+    "s": 12,
+    "t": 10,
+    "u": 23,
+    "w": 18,
+    "x": 34,
+    "y": 24,
+    "z": 30,
+    "|": 4,
+    "ă": 5,
+    "ĕ": 15,
+    "š": 25,
+    "ž": 39,
+    "ǝ": 37,
+    "ɣ": 17,
+    "ḍ": 29,
+    "ṣ": 38,
+    "ṭ": 35,
+    "ẓ": 36
+  },
+  "tat": {
+    "-": 35,
+    "1": 56,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 44,
+    "e": 45,
+    "o": 47,
+    "x": 49,
+    "|": 4,
+    "μ": 53,
+    "а": 5,
+    "б": 15,
+    "в": 33,
+    "г": 17,
+    "д": 18,
+    "е": 6,
+    "ж": 43,
+    "з": 22,
+    "и": 14,
+    "й": 23,
+    "к": 12,
+    "л": 10,
+    "м": 16,
+    "н": 7,
+    "о": 28,
+    "п": 25,
+    "р": 8,
+    "с": 19,
+    "т": 13,
+    "у": 20,
+    "ф": 39,
+    "х": 31,
+    "ц": 41,
+    "ч": 24,
+    "ш": 21,
+    "щ": 48,
+    "ъ": 40,
+    "ы": 11,
+    "ь": 38,
+    "э": 34,
+    "ю": 37,
+    "я": 30,
+    "ё": 50,
+    "ђ": 54,
+    "њ": 51,
+    "ћ": 52,
+    "ѳ": 55,
+    "җ": 32,
+    "ң": 26,
+    "ү": 27,
+    "һ": 36,
+    "ә": 9,
+    "ө": 29,
+    "–": 46,
+    "—": 42
+  },
+  "tav": {
+    "-": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 9,
+    "d": 28,
+    "e": 10,
+    "f": 38,
+    "g": 23,
+    "h": 39,
+    "i": 6,
+    "j": 15,
+    "k": 45,
+    "l": 31,
+    "m": 16,
+    "n": 12,
+    "o": 8,
+    "p": 17,
+    "q": 22,
+    "r": 7,
+    "s": 27,
+    "t": 14,
+    "u": 13,
+    "v": 40,
+    "w": 25,
+    "y": 20,
+    "z": 43,
+    "|": 4,
+    "á": 26,
+    "ã": 19,
+    "é": 36,
+    "í": 32,
+    "ñ": 24,
+    "ó": 37,
+    "õ": 29,
+    "ú": 42,
+    "ĩ": 21,
+    "ũ": 35,
+    "ʉ": 11,
+    "́": 41,
+    "̃": 34,
+    "ẽ": 30,
+    "—": 33
+  },
+  "tbc": {
+    "'": 31,
+    "-": 38,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 35,
+    "4": 32,
+    "5": 34,
+    "6": 36,
+    "7": 33,
+    "8": 39,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 10,
+    "e": 11,
+    "f": 24,
+    "g": 15,
+    "h": 26,
+    "i": 6,
+    "j": 25,
+    "k": 18,
+    "l": 13,
+    "m": 12,
+    "n": 7,
+    "o": 9,
+    "p": 19,
+    "r": 20,
+    "s": 17,
+    "t": 14,
+    "u": 8,
+    "v": 27,
+    "w": 23,
+    "y": 22,
+    "|": 4,
+    "ŋ": 16
+  },
+  "tbg": {
+    "-": 35,
+    "0": 25,
+    "1": 29,
+    "2": 31,
+    "3": 37,
+    "4": 33,
+    "5": 36,
+    "6": 38,
+    "7": 32,
+    "8": 40,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "d": 24,
+    "e": 13,
+    "f": 30,
+    "g": 28,
+    "h": 17,
+    "i": 6,
+    "j": 34,
+    "k": 14,
+    "l": 21,
+    "m": 16,
+    "n": 10,
+    "o": 11,
+    "p": 19,
+    "q": 15,
+    "r": 8,
+    "s": 20,
+    "t": 9,
+    "u": 12,
+    "v": 7,
+    "y": 42,
+    "|": 5,
+    "­": 18,
+    "á": 22,
+    "˻": 26,
+    "˼": 27,
+    "—": 39
+  },
+  "tbk": {
+    "'": 32,
+    "-": 24,
+    "0": 46,
+    "1": 34,
+    "2": 38,
+    "3": 39,
+    "4": 40,
+    "5": 41,
+    "6": 42,
+    "7": 43,
+    "8": 44,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 26,
+    "d": 18,
+    "e": 15,
+    "f": 28,
+    "g": 7,
+    "h": 27,
+    "i": 9,
+    "j": 25,
+    "k": 16,
+    "l": 19,
+    "m": 11,
+    "n": 6,
+    "o": 23,
+    "p": 13,
+    "q": 31,
+    "r": 17,
+    "s": 14,
+    "t": 10,
+    "u": 8,
+    "v": 30,
+    "w": 22,
+    "x": 33,
+    "y": 12,
+    "z": 29,
+    "|": 5,
+    "ë": 37,
+    "̓": 36,
+    "̠": 35,
+    "̱": 20
+  },
+  "tbl": {
+    "'": 36,
+    "-": 35,
+    "6": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 18,
+    "c": 38,
+    "d": 15,
+    "e": 5,
+    "f": 27,
+    "g": 16,
+    "h": 20,
+    "i": 21,
+    "j": 29,
+    "k": 13,
+    "l": 9,
+    "m": 7,
+    "n": 6,
+    "o": 17,
+    "p": 34,
+    "r": 30,
+    "s": 19,
+    "t": 12,
+    "u": 8,
+    "w": 22,
+    "y": 14,
+    "|": 4,
+    "à": 26,
+    "è": 23,
+    "é": 24,
+    "ê": 31,
+    "ì": 33,
+    "ò": 28,
+    "ó": 11,
+    "ô": 25,
+    "ù": 32
+  },
+  "tby": {
+    "'": 11,
+    "-": 23,
+    "0": 29,
+    "1": 34,
+    "2": 30,
+    "3": 36,
+    "4": 35,
+    "5": 37,
+    "6": 39,
+    "7": 33,
+    "8": 31,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 28,
+    "d": 16,
+    "e": 9,
+    "f": 27,
+    "g": 12,
+    "h": 26,
+    "i": 7,
+    "j": 25,
+    "k": 13,
+    "l": 21,
+    "m": 14,
+    "n": 8,
+    "o": 6,
+    "p": 24,
+    "r": 19,
+    "s": 15,
+    "t": 17,
+    "u": 10,
+    "w": 18,
+    "y": 20,
+    "z": 32,
+    "|": 4,
+    "–": 38
+  },
+  "tbz": {
+    "-": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 41,
+    "b": 15,
+    "c": 33,
+    "d": 14,
+    "e": 22,
+    "f": 35,
+    "h": 36,
+    "i": 9,
+    "k": 7,
+    "m": 12,
+    "n": 8,
+    "o": 20,
+    "p": 25,
+    "r": 28,
+    "s": 24,
+    "t": 11,
+    "u": 17,
+    "v": 47,
+    "w": 30,
+    "y": 18,
+    "|": 4,
+    "á": 45,
+    "ã": 44,
+    "è": 31,
+    "é": 37,
+    "ì": 19,
+    "í": 23,
+    "ò": 26,
+    "ó": 29,
+    "ù": 27,
+    "ú": 32,
+    "ĩ": 34,
+    "ń": 40,
+    "ũ": 38,
+    "ǹ": 39,
+    "ɑ": 10,
+    "ɔ": 16,
+    "ɛ": 5,
+    "̀": 6,
+    "́": 13,
+    "̃": 21,
+    "ḿ": 43,
+    "ṹ": 42
+  },
+  "tca": {
+    "'": 58,
+    "-": 56,
+    "0": 43,
+    "1": 46,
+    "2": 52,
+    "3": 62,
+    "4": 54,
+    "5": 57,
+    "6": 59,
+    "7": 48,
+    "8": 61,
+    "9": 63,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 37,
+    "c": 13,
+    "d": 27,
+    "e": 14,
+    "f": 55,
+    "g": 9,
+    "h": 19,
+    "i": 16,
+    "k": 64,
+    "l": 47,
+    "m": 15,
+    "n": 8,
+    "o": 24,
+    "p": 21,
+    "q": 41,
+    "r": 10,
+    "s": 38,
+    "t": 17,
+    "u": 11,
+    "w": 25,
+    "x": 7,
+    "y": 20,
+    "z": 53,
+    "|": 4,
+    "á": 29,
+    "ã": 22,
+    "é": 33,
+    "í": 30,
+    "ñ": 28,
+    "ó": 39,
+    "õ": 36,
+    "ú": 35,
+    "ü": 6,
+    "ĩ": 26,
+    "ũ": 34,
+    "ǘ": 50,
+    "́": 32,
+    "̃": 12,
+    "̱": 23,
+    "ḇ": 49,
+    "ḏ": 44,
+    "ṉ": 42,
+    "ṍ": 45,
+    "ṯ": 40,
+    "ṹ": 51,
+    "ẖ": 60,
+    "ẽ": 18,
+    "—": 31
+  },
+  "tcc": {
+    "'": 23,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 25,
+    "d": 10,
+    "e": 6,
+    "f": 27,
+    "g": 8,
+    "h": 12,
+    "i": 7,
+    "j": 24,
+    "k": 20,
+    "l": 22,
+    "m": 19,
+    "n": 9,
+    "o": 15,
+    "p": 26,
+    "r": 21,
+    "s": 11,
+    "t": 17,
+    "u": 13,
+    "w": 14,
+    "y": 16,
+    "|": 5
+  },
+  "tcs": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 12,
+    "e": 7,
+    "g": 19,
+    "i": 6,
+    "k": 18,
+    "l": 13,
+    "m": 10,
+    "n": 8,
+    "o": 9,
+    "p": 11,
+    "r": 21,
+    "s": 15,
+    "t": 16,
+    "u": 17,
+    "w": 20,
+    "y": 23,
+    "z": 22,
+    "|": 4
+  },
+  "tcz": {
+    "'": 28,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 43,
+    "a": 5,
+    "b": 27,
+    "c": 18,
+    "d": 22,
+    "e": 17,
+    "f": 36,
+    "g": 8,
+    "h": 9,
+    "i": 7,
+    "j": 31,
+    "k": 13,
+    "l": 16,
+    "m": 12,
+    "n": 6,
+    "o": 23,
+    "p": 20,
+    "r": 14,
+    "s": 19,
+    "t": 11,
+    "u": 10,
+    "v": 25,
+    "w": 15,
+    "z": 24,
+    "|": 4,
+    "á": 35,
+    "â": 21,
+    "ä": 45,
+    "é": 39,
+    "ê": 26,
+    "í": 38,
+    "î": 29,
+    "ó": 41,
+    "û": 32,
+    "ā": 33,
+    "ē": 37,
+    "ī": 40,
+    "ō": 44,
+    "ū": 34,
+    "ȇ": 42
+  },
+  "tdj": {
+    "'": 29,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 27,
+    "d": 26,
+    "e": 9,
+    "g": 16,
+    "h": 23,
+    "i": 6,
+    "j": 21,
+    "k": 20,
+    "l": 15,
+    "m": 13,
+    "n": 7,
+    "o": 8,
+    "p": 18,
+    "r": 17,
+    "s": 12,
+    "t": 11,
+    "u": 10,
+    "v": 25,
+    "w": 28,
+    "y": 19,
+    "|": 5,
+    "ꞌ": 14
+  },
+  "ted": {
+    "'": 5,
+    "1": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 19,
+    "c": 31,
+    "d": 25,
+    "e": 12,
+    "f": 33,
+    "g": 27,
+    "h": 23,
+    "i": 17,
+    "j": 32,
+    "k": 18,
+    "l": 14,
+    "m": 20,
+    "n": 6,
+    "o": 21,
+    "p": 29,
+    "r": 30,
+    "s": 28,
+    "t": 22,
+    "u": 16,
+    "w": 26,
+    "y": 15,
+    "|": 4,
+    "ŋ": 34,
+    "ɔ": 11,
+    "ɛ": 10,
+    "ɩ": 13,
+    "ʋ": 8,
+    "‐": 9,
+    "꞊": 24
+  },
+  "tee": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 9,
+    "d": 26,
+    "e": 25,
+    "f": 31,
+    "g": 30,
+    "h": 12,
+    "i": 6,
+    "j": 13,
+    "k": 17,
+    "l": 14,
+    "m": 19,
+    "n": 8,
+    "o": 20,
+    "p": 18,
+    "q": 23,
+    "r": 24,
+    "s": 15,
+    "t": 10,
+    "u": 7,
+    "v": 21,
+    "x": 16,
+    "y": 22,
+    "z": 35,
+    "|": 5,
+    "á": 34,
+    "é": 29,
+    "í": 32,
+    "ñ": 36,
+    "ó": 33,
+    "ú": 27,
+    "ʼ": 11
+  },
+  "tel": {
+    "!": 118,
+    "\"": 48,
+    "$": 106,
+    "%": 110,
+    "'": 101,
+    "+": 114,
+    ",": 92,
+    "-": 75,
+    ".": 59,
+    "/": 95,
+    "0": 49,
+    "1": 51,
+    "2": 56,
+    "3": 68,
+    "4": 72,
+    "5": 71,
+    "6": 77,
+    "7": 82,
+    "8": 74,
+    "9": 66,
+    ":": 96,
+    ";": 108,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 120,
+    "a": 54,
+    "b": 88,
+    "c": 78,
+    "d": 86,
+    "e": 61,
+    "f": 89,
+    "g": 84,
+    "h": 80,
+    "i": 62,
+    "j": 103,
+    "k": 91,
+    "l": 73,
+    "m": 76,
+    "n": 67,
+    "o": 65,
+    "p": 87,
+    "q": 111,
+    "r": 64,
+    "s": 60,
+    "t": 70,
+    "u": 69,
+    "v": 99,
+    "w": 94,
+    "x": 109,
+    "y": 93,
+    "z": 102,
+    "|": 4,
+    "°": 113,
+    "²": 119,
+    "ç": 115,
+    "õ": 112,
+    "ం": 11,
+    "ః": 117,
+    "అ": 28,
+    "ఆ": 41,
+    "ఇ": 40,
+    "ఈ": 52,
+    "ఉ": 33,
+    "ఊ": 97,
+    "ఋ": 116,
+    "ఎ": 44,
+    "ఏ": 55,
+    "ఐ": 81,
+    "ఒ": 47,
+    "ఓ": 79,
+    "ఔ": 107,
+    "క": 13,
+    "ఖ": 57,
+    "గ": 26,
+    "ఘ": 83,
+    "చ": 22,
+    "ఛ": 90,
+    "జ": 31,
+    "ఝ": 123,
+    "ఞ": 98,
+    "ట": 21,
+    "ఠ": 100,
+    "డ": 23,
+    "ఢ": 104,
+    "ణ": 39,
+    "త": 14,
+    "థ": 46,
+    "ద": 19,
+    "ధ": 35,
+    "న": 10,
+    "ప": 15,
+    "ఫ": 45,
+    "బ": 30,
+    "భ": 42,
+    "మ": 20,
+    "య": 18,
+    "ర": 9,
+    "ఱ": 122,
+    "ల": 12,
+    "ళ": 50,
+    "వ": 16,
+    "శ": 38,
+    "ష": 37,
+    "స": 17,
+    "హ": 43,
+    "ా": 7,
+    "ి": 6,
+    "ీ": 27,
+    "ు": 8,
+    "ూ": 34,
+    "ృ": 53,
+    "ౄ": 124,
+    "ె": 29,
+    "ే": 24,
+    "ై": 32,
+    "ొ": 36,
+    "ో": 25,
+    "ౌ": 58,
+    "్": 5,
+    "౦": 63,
+    "​": 105,
+    "‌": 85,
+    "–": 121
+  },
+  "tem": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 32,
+    "d": 29,
+    "e": 19,
+    "f": 27,
+    "g": 31,
+    "h": 23,
+    "i": 8,
+    "k": 6,
+    "l": 24,
+    "m": 11,
+    "n": 14,
+    "o": 22,
+    "p": 25,
+    "r": 13,
+    "s": 17,
+    "t": 9,
+    "u": 21,
+    "w": 26,
+    "y": 18,
+    "|": 4,
+    "ŋ": 10,
+    "ǝ": 30,
+    "ȧ": 16,
+    "ɔ": 7,
+    "ə": 12,
+    "ɛ": 15
+  },
+  "teo": {
+    "'": 26,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 24,
+    "d": 17,
+    "e": 8,
+    "g": 25,
+    "i": 6,
+    "j": 22,
+    "k": 7,
+    "l": 15,
+    "m": 16,
+    "n": 10,
+    "o": 9,
+    "p": 19,
+    "r": 13,
+    "s": 14,
+    "t": 11,
+    "u": 12,
+    "w": 23,
+    "y": 18,
+    "|": 5,
+    "ŋ": 20,
+    "–": 29,
+    "—": 28
+  },
+  "ter": {
+    "'": 19,
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 38,
+    "d": 31,
+    "e": 7,
+    "f": 40,
+    "g": 34,
+    "h": 15,
+    "i": 8,
+    "j": 29,
+    "k": 6,
+    "l": 33,
+    "m": 20,
+    "n": 9,
+    "o": 5,
+    "p": 13,
+    "r": 27,
+    "s": 24,
+    "t": 11,
+    "u": 12,
+    "v": 14,
+    "x": 17,
+    "y": 16,
+    "z": 37,
+    "|": 4,
+    "á": 21,
+    "â": 26,
+    "é": 25,
+    "ê": 28,
+    "í": 22,
+    "î": 39,
+    "ó": 18,
+    "ô": 32,
+    "ú": 23,
+    "û": 30,
+    "‍": 36,
+    "—": 35
+  },
+  "tes": {
+    "'": 25,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 27,
+    "d": 17,
+    "e": 7,
+    "f": 28,
+    "g": 9,
+    "h": 22,
+    "i": 8,
+    "j": 24,
+    "k": 10,
+    "l": 20,
+    "m": 15,
+    "n": 6,
+    "o": 14,
+    "p": 18,
+    "r": 13,
+    "s": 12,
+    "t": 16,
+    "u": 11,
+    "w": 19,
+    "y": 23,
+    "|": 4
+  },
+  "tew": {
+    "'": 49,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 29,
+    "c": 40,
+    "d": 21,
+    "e": 10,
+    "f": 47,
+    "g": 26,
+    "h": 11,
+    "i": 6,
+    "j": 34,
+    "k": 24,
+    "l": 43,
+    "m": 25,
+    "n": 5,
+    "o": 16,
+    "p": 23,
+    "q": 48,
+    "r": 41,
+    "s": 22,
+    "t": 18,
+    "u": 19,
+    "v": 27,
+    "w": 12,
+    "x": 46,
+    "y": 17,
+    "z": 45,
+    "|": 4,
+    "á": 13,
+    "â": 28,
+    "ä": 20,
+    "é": 35,
+    "ê": 42,
+    "í": 9,
+    "î": 38,
+    "ó": 30,
+    "ô": 36,
+    "ú": 33,
+    "û": 39,
+    "ˀ": 8,
+    "́": 37,
+    "̂": 44,
+    "̖": 15,
+    "ḏ": 14,
+    "—": 50,
+    "ꞌ": 31
+  },
+  "tex": {
+    "'": 32,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 22,
+    "c": 15,
+    "d": 19,
+    "e": 9,
+    "g": 13,
+    "h": 27,
+    "i": 8,
+    "j": 26,
+    "k": 11,
+    "l": 16,
+    "m": 20,
+    "n": 7,
+    "o": 10,
+    "p": 29,
+    "r": 17,
+    "s": 25,
+    "t": 12,
+    "u": 14,
+    "v": 24,
+    "w": 23,
+    "y": 21,
+    "z": 18,
+    "|": 4,
+    "á": 33,
+    "í": 34,
+    "ó": 28,
+    "̱": 5,
+    "ṯ": 31
+  },
+  "tfr": {
+    "'": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 34,
+    "d": 23,
+    "e": 10,
+    "f": 40,
+    "g": 8,
+    "h": 21,
+    "i": 15,
+    "j": 9,
+    "k": 7,
+    "l": 36,
+    "m": 24,
+    "n": 12,
+    "o": 5,
+    "p": 22,
+    "q": 44,
+    "r": 13,
+    "s": 14,
+    "t": 20,
+    "u": 27,
+    "v": 43,
+    "w": 19,
+    "x": 48,
+    "y": 18,
+    "z": 35,
+    "|": 4,
+    "à": 50,
+    "á": 42,
+    "ã": 31,
+    "ä": 28,
+    "è": 47,
+    "é": 38,
+    "ê": 33,
+    "ë": 17,
+    "í": 37,
+    "î": 29,
+    "ï": 53,
+    "ð": 26,
+    "ñ": 54,
+    "ò": 45,
+    "ó": 41,
+    "õ": 30,
+    "ö": 25,
+    "ø": 46,
+    "ú": 32,
+    "û": 39,
+    "ü": 52,
+    "þ": 51,
+    "š": 11
+  },
+  "tgj": {
+    "'": 29,
+    "-": 28,
+    "0": 30,
+    "1": 33,
+    "2": 36,
+    "4": 31,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 16,
+    "c": 26,
+    "d": 17,
+    "e": 25,
+    "f": 34,
+    "g": 9,
+    "h": 23,
+    "i": 11,
+    "j": 24,
+    "k": 12,
+    "l": 14,
+    "m": 13,
+    "n": 6,
+    "o": 8,
+    "p": 22,
+    "r": 15,
+    "s": 21,
+    "t": 20,
+    "u": 10,
+    "v": 5,
+    "w": 18,
+    "y": 19,
+    "z": 35,
+    "|": 4,
+    "—": 27
+  },
+  "tgk": {
+    "!": 99,
+    "\"": 78,
+    "%": 86,
+    "'": 84,
+    "+": 96,
+    ",": 76,
+    "-": 40,
+    ".": 64,
+    "/": 71,
+    "0": 36,
+    "1": 38,
+    "2": 41,
+    "3": 54,
+    "4": 46,
+    "5": 47,
+    "6": 53,
+    "7": 62,
+    "8": 56,
+    "9": 44,
+    ":": 70,
+    ";": 94,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 101,
+    "[": 102,
+    "\\": 97,
+    "]": 103,
+    "a": 42,
+    "b": 72,
+    "c": 60,
+    "d": 65,
+    "e": 43,
+    "f": 77,
+    "g": 67,
+    "h": 69,
+    "i": 55,
+    "j": 83,
+    "k": 75,
+    "l": 58,
+    "m": 66,
+    "n": 48,
+    "o": 50,
+    "p": 59,
+    "q": 90,
+    "r": 52,
+    "s": 49,
+    "t": 51,
+    "u": 68,
+    "v": 74,
+    "w": 73,
+    "x": 81,
+    "y": 79,
+    "z": 87,
+    "|": 4,
+    "«": 63,
+    "­": 100,
+    "°": 89,
+    "²": 98,
+    "»": 61,
+    "õ": 93,
+    "а": 5,
+    "б": 13,
+    "в": 19,
+    "г": 23,
+    "д": 10,
+    "е": 18,
+    "ж": 57,
+    "з": 22,
+    "и": 6,
+    "й": 31,
+    "к": 17,
+    "л": 20,
+    "м": 11,
+    "н": 9,
+    "о": 7,
+    "п": 30,
+    "р": 8,
+    "с": 15,
+    "т": 12,
+    "у": 14,
+    "ф": 24,
+    "х": 26,
+    "ц": 95,
+    "ч": 35,
+    "ш": 21,
+    "щ": 91,
+    "ъ": 34,
+    "ь": 85,
+    "э": 39,
+    "ю": 45,
+    "я": 27,
+    "ё": 32,
+    "ғ": 37,
+    "қ": 25,
+    "ҳ": 16,
+    "ҷ": 28,
+    "ӣ": 29,
+    "ӯ": 33,
+    "​": 80,
+    "‐": 105,
+    "‑": 82,
+    "–": 88,
+    "—": 104,
+    "№": 92
+  },
+  "tgl": {
+    "!": 61,
+    "\"": 57,
+    "$": 52,
+    "%": 53,
+    "&": 56,
+    "'": 31,
+    "+": 67,
+    ",": 43,
+    "-": 24,
+    ".": 34,
+    "/": 46,
+    "0": 26,
+    "1": 27,
+    "2": 30,
+    "3": 40,
+    "4": 38,
+    "5": 35,
+    "6": 36,
+    "7": 41,
+    "8": 39,
+    "9": 32,
+    ":": 45,
+    ";": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 25,
+    "d": 22,
+    "e": 20,
+    "f": 28,
+    "g": 7,
+    "h": 21,
+    "i": 8,
+    "j": 33,
+    "k": 15,
+    "l": 13,
+    "m": 12,
+    "n": 6,
+    "o": 11,
+    "p": 14,
+    "q": 44,
+    "r": 18,
+    "s": 9,
+    "t": 10,
+    "u": 16,
+    "v": 29,
+    "w": 23,
+    "x": 42,
+    "y": 17,
+    "z": 37,
+    "|": 5,
+    "¡": 73,
+    "£": 63,
+    "¥": 60,
+    "§": 66,
+    "­": 54,
+    "°": 68,
+    "µ": 70,
+    "¶": 71,
+    "º": 72,
+    "¼": 65,
+    "á": 50,
+    "â": 47,
+    "ã": 48,
+    "í": 62,
+    "î": 81,
+    "ñ": 80,
+    "ó": 78,
+    "ô": 77,
+    "ú": 79,
+    "œ": 64,
+    "ʼ": 74,
+    "˜": 58,
+    "–": 75,
+    "—": 76,
+    "”": 51,
+    "•": 69,
+    "€": 49,
+    "™": 59
+  },
+  "tgo": {
+    "'": 32,
+    "-": 35,
+    "0": 28,
+    "1": 30,
+    "2": 31,
+    "6": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "d": 25,
+    "e": 6,
+    "f": 33,
+    "g": 9,
+    "h": 8,
+    "i": 7,
+    "j": 24,
+    "k": 18,
+    "l": 22,
+    "m": 13,
+    "n": 11,
+    "o": 10,
+    "p": 27,
+    "r": 12,
+    "s": 26,
+    "t": 14,
+    "u": 16,
+    "v": 19,
+    "w": 17,
+    "y": 15,
+    "|": 4,
+    "ŋ": 21,
+    "ɨ": 20,
+    "—": 29
+  },
+  "tgp": {
+    "'": 31,
+    "-": 45,
+    "0": 36,
+    "1": 27,
+    "2": 29,
+    "3": 33,
+    "4": 32,
+    "5": 37,
+    "6": 34,
+    "7": 40,
+    "8": 39,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 25,
+    "d": 23,
+    "e": 9,
+    "f": 35,
+    "g": 21,
+    "h": 15,
+    "i": 6,
+    "j": 22,
+    "k": 20,
+    "l": 12,
+    "m": 11,
+    "n": 8,
+    "o": 7,
+    "p": 18,
+    "q": 44,
+    "r": 10,
+    "s": 16,
+    "t": 13,
+    "u": 14,
+    "v": 17,
+    "w": 26,
+    "x": 41,
+    "y": 28,
+    "z": 30,
+    "|": 4,
+    "̃": 19,
+    "–": 42,
+    "—": 43
+  },
+  "tha": {
+    "\"": 124,
+    "%": 110,
+    "&": 127,
+    "'": 105,
+    "+": 126,
+    ",": 93,
+    "-": 96,
+    ".": 68,
+    "/": 102,
+    "0": 53,
+    "1": 57,
+    "2": 63,
+    "3": 78,
+    "4": 76,
+    "5": 70,
+    "6": 83,
+    "7": 89,
+    "8": 86,
+    "9": 79,
+    ":": 121,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 119,
+    "]": 120,
+    "a": 52,
+    "b": 90,
+    "c": 73,
+    "d": 80,
+    "e": 56,
+    "f": 91,
+    "g": 85,
+    "h": 77,
+    "i": 65,
+    "j": 100,
+    "k": 87,
+    "l": 67,
+    "m": 82,
+    "n": 62,
+    "o": 59,
+    "p": 81,
+    "q": 108,
+    "r": 61,
+    "s": 64,
+    "t": 66,
+    "u": 75,
+    "v": 98,
+    "w": 95,
+    "x": 104,
+    "y": 94,
+    "z": 106,
+    "°": 114,
+    "á": 111,
+    "ç": 130,
+    "é": 122,
+    "í": 117,
+    "ó": 125,
+    "õ": 115,
+    "ú": 116,
+    "ü": 113,
+    "ł": 131,
+    "ก": 8,
+    "ข": 32,
+    "ค": 28,
+    "ฆ": 101,
+    "ง": 10,
+    "จ": 30,
+    "ฉ": 69,
+    "ช": 37,
+    "ซ": 45,
+    "ฌ": 129,
+    "ญ": 51,
+    "ฎ": 84,
+    "ฏ": 88,
+    "ฐ": 60,
+    "ฑ": 107,
+    "ฒ": 92,
+    "ณ": 48,
+    "ด": 20,
+    "ต": 22,
+    "ถ": 44,
+    "ท": 19,
+    "ธ": 54,
+    "น": 5,
+    "บ": 25,
+    "ป": 27,
+    "ผ": 47,
+    "ฝ": 71,
+    "พ": 35,
+    "ฟ": 55,
+    "ภ": 49,
+    "ม": 13,
+    "ย": 17,
+    "ร": 7,
+    "ฤ": 74,
+    "ล": 18,
+    "ว": 16,
+    "ศ": 46,
+    "ษ": 50,
+    "ส": 21,
+    "ห": 24,
+    "ฬ": 99,
+    "อ": 9,
+    "ฮ": 72,
+    "ฯ": 109,
+    "ะ": 26,
+    "ั": 15,
+    "า": 4,
+    "ำ": 41,
+    "ิ": 23,
+    "ี": 14,
+    "ึ": 43,
+    "ื": 33,
+    "ุ": 40,
+    "ู": 38,
+    "เ": 11,
+    "แ": 29,
+    "โ": 42,
+    "ใ": 31,
+    "ไ": 34,
+    "ๅ": 132,
+    "ๆ": 58,
+    "็": 39,
+    "่": 6,
+    "้": 12,
+    "๊": 97,
+    "๋": 103,
+    "์": 36,
+    "ํ": 112,
+    "​": 118,
+    "–": 128,
+    "”": 123
+  },
+  "thk": {
+    "'": 24,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 15,
+    "c": 25,
+    "d": 22,
+    "e": 12,
+    "f": 29,
+    "g": 14,
+    "h": 20,
+    "i": 10,
+    "j": 26,
+    "k": 13,
+    "l": 30,
+    "m": 16,
+    "n": 6,
+    "o": 19,
+    "p": 27,
+    "r": 9,
+    "s": 23,
+    "t": 11,
+    "u": 18,
+    "w": 17,
+    "y": 21,
+    "|": 5,
+    "î": 8,
+    "û": 7
+  },
+  "thl": {
+    "'": 64,
+    "-": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 25,
+    "ं": 58,
+    "ः": 62,
+    "अ": 36,
+    "आ": 38,
+    "इ": 32,
+    "ई": 55,
+    "उ": 37,
+    "ऊ": 51,
+    "ऋ": 65,
+    "ए": 46,
+    "ऐ": 66,
+    "ओ": 33,
+    "औ": 57,
+    "क": 8,
+    "ख": 35,
+    "ग": 30,
+    "घ": 49,
+    "ङ": 56,
+    "च": 39,
+    "छ": 44,
+    "ज": 28,
+    "झ": 52,
+    "ञ": 59,
+    "ट": 26,
+    "ठ": 34,
+    "ड": 47,
+    "ढ": 61,
+    "ण": 50,
+    "त": 21,
+    "थ": 48,
+    "द": 29,
+    "ध": 43,
+    "न": 10,
+    "प": 17,
+    "फ": 42,
+    "ब": 23,
+    "भ": 41,
+    "म": 11,
+    "य": 24,
+    "र": 6,
+    "ल": 13,
+    "व": 20,
+    "श": 31,
+    "ष": 45,
+    "स": 15,
+    "ह": 9,
+    "ऽ": 63,
+    "ा": 5,
+    "ि": 18,
+    "ी": 22,
+    "ु": 16,
+    "ू": 40,
+    "ृ": 60,
+    "े": 14,
+    "ै": 19,
+    "ो": 27,
+    "ौ": 53,
+    "्": 7,
+    "‍": 12
+  },
+  "tih": {
+    "'": 17,
+    "-": 24,
+    "0": 29,
+    "1": 32,
+    "2": 34,
+    "4": 30,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 28,
+    "d": 21,
+    "e": 26,
+    "g": 13,
+    "h": 25,
+    "i": 8,
+    "j": 22,
+    "k": 11,
+    "l": 14,
+    "m": 12,
+    "n": 6,
+    "o": 7,
+    "p": 18,
+    "r": 10,
+    "s": 16,
+    "t": 15,
+    "u": 9,
+    "w": 23,
+    "y": 20,
+    "z": 31,
+    "|": 5,
+    "–": 27
+  },
+  "tik": {
+    "-": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 18,
+    "b": 23,
+    "c": 30,
+    "d": 34,
+    "e": 20,
+    "f": 31,
+    "g": 39,
+    "h": 28,
+    "i": 7,
+    "j": 43,
+    "k": 14,
+    "l": 13,
+    "m": 8,
+    "n": 5,
+    "o": 27,
+    "p": 40,
+    "r": 45,
+    "s": 15,
+    "t": 21,
+    "u": 17,
+    "v": 37,
+    "w": 9,
+    "y": 10,
+    "z": 29,
+    "|": 4,
+    "à": 22,
+    "â": 46,
+    "æ": 24,
+    "è": 42,
+    "ê": 50,
+    "ì": 19,
+    "î": 49,
+    "ò": 38,
+    "ô": 55,
+    "ù": 25,
+    "û": 52,
+    "ě": 53,
+    "ŋ": 36,
+    "ǎ": 41,
+    "ǐ": 48,
+    "ǒ": 54,
+    "ǔ": 51,
+    "ǹ": 33,
+    "ɓ": 11,
+    "ɔ": 26,
+    "ɗ": 35,
+    "ɛ": 6,
+    "ʼ": 16,
+    "̀": 12,
+    "̂": 44,
+    "̌": 32
+  },
+  "tir": {
+    "-": 215,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "h": 224,
+    "|": 4,
+    "ሀ": 186,
+    "ሁ": 122,
+    "ሂ": 172,
+    "ሃ": 78,
+    "ሄ": 181,
+    "ህ": 86,
+    "ሆ": 138,
+    "ለ": 22,
+    "ሉ": 40,
+    "ሊ": 73,
+    "ላ": 63,
+    "ሌ": 141,
+    "ል": 14,
+    "ሎ": 45,
+    "ሐ": 133,
+    "ሑ": 128,
+    "ሒ": 157,
+    "ሓ": 49,
+    "ሔ": 67,
+    "ሕ": 35,
+    "ሖ": 180,
+    "መ": 20,
+    "ሙ": 96,
+    "ሚ": 105,
+    "ማ": 34,
+    "ሜ": 166,
+    "ም": 6,
+    "ሞ": 75,
+    "ሠ": 109,
+    "ሡ": 204,
+    "ሢ": 170,
+    "ሣ": 150,
+    "ሥ": 77,
+    "ረ": 32,
+    "ሩ": 65,
+    "ሪ": 74,
+    "ራ": 62,
+    "ሬ": 177,
+    "ር": 13,
+    "ሮ": 89,
+    "ሰ": 28,
+    "ሱ": 60,
+    "ሲ": 123,
+    "ሳ": 61,
+    "ሴ": 160,
+    "ስ": 12,
+    "ሶ": 146,
+    "ሸ": 140,
+    "ሹ": 198,
+    "ሺ": 218,
+    "ሻ": 159,
+    "ሼ": 228,
+    "ሽ": 99,
+    "ሾ": 206,
+    "ቀ": 101,
+    "ቁ": 187,
+    "ቂ": 117,
+    "ቃ": 121,
+    "ቄ": 178,
+    "ቅ": 83,
+    "ቆ": 149,
+    "ቈ": 193,
+    "ቊ": 175,
+    "ቋ": 171,
+    "ቍ": 222,
+    "ቐ": 79,
+    "ቑ": 126,
+    "ቒ": 183,
+    "ቓ": 113,
+    "ቕ": 66,
+    "ቖ": 143,
+    "ቛ": 189,
+    "ቝ": 162,
+    "በ": 19,
+    "ቡ": 106,
+    "ቢ": 91,
+    "ባ": 52,
+    "ቤ": 115,
+    "ብ": 8,
+    "ቦ": 104,
+    "ተ": 16,
+    "ቱ": 81,
+    "ቲ": 21,
+    "ታ": 47,
+    "ቴ": 185,
+    "ት": 11,
+    "ቶ": 38,
+    "ቹ": 226,
+    "ቺ": 230,
+    "ች": 213,
+    "ቾ": 221,
+    "ኀ": 197,
+    "ኁ": 173,
+    "ኂ": 165,
+    "ኃ": 69,
+    "ኅ": 71,
+    "ኆ": 216,
+    "ነ": 18,
+    "ኑ": 88,
+    "ኒ": 93,
+    "ና": 15,
+    "ኔ": 179,
+    "ን": 5,
+    "ኖ": 118,
+    "ኛ": 174,
+    "ኝ": 227,
+    "አ": 100,
+    "ኡ": 64,
+    "ኢ": 46,
+    "ኣ": 9,
+    "ኤ": 131,
+    "እ": 7,
+    "ኦ": 97,
+    "ከ": 27,
+    "ኩ": 42,
+    "ኪ": 129,
+    "ካ": 44,
+    "ኬ": 203,
+    "ክ": 29,
+    "ኮ": 98,
+    "ኰ": 191,
+    "ኲ": 108,
+    "ኳ": 116,
+    "ኴ": 194,
+    "ኸ": 37,
+    "ኹ": 51,
+    "ኺ": 156,
+    "ኻ": 53,
+    "ኼ": 208,
+    "ኽ": 33,
+    "ኾ": 80,
+    "ዀ": 209,
+    "ዂ": 95,
+    "ዃ": 161,
+    "ዄ": 201,
+    "ዅ": 199,
+    "ወ": 43,
+    "ዉ": 87,
+    "ዊ": 110,
+    "ዋ": 82,
+    "ዌ": 212,
+    "ው": 30,
+    "ዎ": 54,
+    "ዐ": 127,
+    "ዑ": 102,
+    "ዒ": 152,
+    "ዓ": 26,
+    "ዕ": 41,
+    "ዖ": 135,
+    "ዘ": 25,
+    "ዙ": 55,
+    "ዚ": 57,
+    "ዛ": 70,
+    "ዜ": 144,
+    "ዝ": 17,
+    "ዞ": 147,
+    "የ": 39,
+    "ዩ": 50,
+    "ዪ": 168,
+    "ያ": 59,
+    "ይ": 10,
+    "ዮ": 76,
+    "ደ": 31,
+    "ዱ": 92,
+    "ዲ": 85,
+    "ዳ": 84,
+    "ዴ": 200,
+    "ድ": 23,
+    "ዶ": 111,
+    "ጀ": 148,
+    "ጃ": 188,
+    "ጅ": 205,
+    "ገ": 36,
+    "ጉ": 154,
+    "ጊ": 107,
+    "ጋ": 94,
+    "ጌ": 145,
+    "ግ": 24,
+    "ጎ": 176,
+    "ጐ": 90,
+    "ጓ": 163,
+    "ጕ": 137,
+    "ጠ": 103,
+    "ጡ": 125,
+    "ጢ": 114,
+    "ጣ": 124,
+    "ጤ": 207,
+    "ጥ": 72,
+    "ጦ": 153,
+    "ጨ": 182,
+    "ጩ": 223,
+    "ጪ": 225,
+    "ጫ": 210,
+    "ጭ": 184,
+    "ጮ": 219,
+    "ጰ": 214,
+    "ጱ": 229,
+    "ጲ": 196,
+    "ጳ": 132,
+    "ጴ": 158,
+    "ጵ": 190,
+    "ጶ": 195,
+    "ጸ": 68,
+    "ጹ": 136,
+    "ጺ": 130,
+    "ጻ": 119,
+    "ጼ": 211,
+    "ጽ": 58,
+    "ጾ": 169,
+    "ፀ": 155,
+    "ፁ": 164,
+    "ፂ": 151,
+    "ፃ": 139,
+    "ፅ": 142,
+    "ፆ": 192,
+    "ፈ": 56,
+    "ፉ": 120,
+    "ፊ": 134,
+    "ፋ": 112,
+    "ፌ": 202,
+    "ፍ": 48,
+    "ፎ": 167,
+    "ፑ": 217,
+    "ፔ": 220
+  },
+  "tkr": {
+    "'": 8,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 30,
+    "d": 24,
+    "e": 6,
+    "f": 39,
+    "g": 21,
+    "h": 13,
+    "i": 9,
+    "k": 17,
+    "l": 12,
+    "m": 16,
+    "n": 7,
+    "o": 27,
+    "p": 32,
+    "q": 18,
+    "r": 20,
+    "s": 14,
+    "t": 28,
+    "u": 15,
+    "v": 25,
+    "x": 22,
+    "y": 11,
+    "z": 33,
+    "|": 4,
+    "ç": 29,
+    "ö": 36,
+    "ü": 35,
+    "ğ": 31,
+    "ı": 10,
+    "ş": 23,
+    "ə": 26,
+    "̇": 38,
+    "–": 34
+  },
+  "tlb": {
+    "'": 28,
+    "-": 25,
+    "0": 36,
+    "1": 39,
+    "2": 41,
+    "4": 38,
+    "6": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 35,
+    "d": 15,
+    "e": 14,
+    "f": 33,
+    "g": 9,
+    "h": 12,
+    "i": 8,
+    "j": 29,
+    "k": 13,
+    "l": 20,
+    "m": 10,
+    "n": 7,
+    "o": 6,
+    "p": 27,
+    "r": 17,
+    "s": 22,
+    "t": 16,
+    "u": 11,
+    "v": 37,
+    "w": 18,
+    "y": 19,
+    "z": 34,
+    "|": 5,
+    "ä": 26,
+    "ë": 24,
+    "ï": 30,
+    "ö": 23,
+    "ü": 32,
+    "̄": 31
+  },
+  "tlj": {
+    "'": 30,
+    "-": 27,
+    "5": 32,
+    "6": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 10,
+    "c": 28,
+    "d": 22,
+    "e": 8,
+    "f": 26,
+    "g": 17,
+    "h": 14,
+    "i": 6,
+    "j": 23,
+    "k": 12,
+    "l": 13,
+    "m": 15,
+    "n": 11,
+    "o": 16,
+    "p": 25,
+    "s": 21,
+    "t": 19,
+    "u": 7,
+    "v": 29,
+    "w": 20,
+    "y": 18,
+    "|": 5,
+    "ʼ": 24,
+    "̱": 9
+  },
+  "tly": {
+    "-": 35,
+    "0": 37,
+    "1": 38,
+    "2": 39,
+    "4": 41,
+    "5": 42,
+    "6": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 15,
+    "b": 11,
+    "c": 32,
+    "d": 9,
+    "e": 13,
+    "f": 33,
+    "h": 22,
+    "i": 8,
+    "j": 31,
+    "k": 18,
+    "l": 23,
+    "m": 12,
+    "n": 7,
+    "o": 6,
+    "p": 30,
+    "q": 28,
+    "r": 16,
+    "s": 20,
+    "t": 17,
+    "u": 24,
+    "v": 21,
+    "x": 27,
+    "y": 14,
+    "z": 26,
+    "|": 4,
+    "ç": 25,
+    "ğ": 29,
+    "ı": 10,
+    "ş": 19,
+    "ə": 5,
+    "̇": 34,
+    "–": 36
+  },
+  "tmc": {
+    "'": 9,
+    "-": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "d": 12,
+    "e": 17,
+    "g": 10,
+    "h": 26,
+    "i": 15,
+    "j": 22,
+    "k": 16,
+    "l": 21,
+    "m": 18,
+    "n": 7,
+    "o": 23,
+    "p": 27,
+    "r": 19,
+    "s": 28,
+    "t": 25,
+    "u": 14,
+    "w": 24,
+    "y": 13,
+    "|": 4,
+    "ɓ": 32,
+    "ɔ": 31,
+    "ɗ": 11,
+    "ɛ": 30,
+    "ɨ": 33,
+    "́": 36,
+    "̰": 34,
+    "ᵻ": 5,
+    "ḭ": 35,
+    "꞊": 29
+  },
+  "tmf": {
+    "'": 17,
+    "0": 30,
+    "1": 31,
+    "2": 33,
+    "3": 38,
+    "4": 35,
+    "5": 36,
+    "6": 39,
+    "7": 34,
+    "8": 41,
+    "9": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 25,
+    "d": 26,
+    "e": 7,
+    "f": 28,
+    "g": 18,
+    "h": 11,
+    "i": 15,
+    "j": 23,
+    "k": 6,
+    "l": 10,
+    "m": 9,
+    "n": 8,
+    "o": 12,
+    "p": 14,
+    "q": 20,
+    "r": 21,
+    "s": 16,
+    "t": 13,
+    "u": 22,
+    "v": 19,
+    "x": 37,
+    "y": 29,
+    "z": 32,
+    "|": 5,
+    "ñ": 42,
+    "—": 24
+  },
+  "tna": {
+    "'": 24,
+    "0": 38,
+    "8": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 14,
+    "d": 13,
+    "e": 7,
+    "f": 28,
+    "g": 29,
+    "h": 17,
+    "i": 8,
+    "j": 12,
+    "k": 36,
+    "l": 23,
+    "m": 15,
+    "n": 9,
+    "o": 20,
+    "p": 16,
+    "q": 21,
+    "r": 19,
+    "s": 11,
+    "t": 10,
+    "u": 6,
+    "v": 33,
+    "x": 35,
+    "y": 22,
+    "z": 34,
+    "|": 5,
+    "á": 31,
+    "é": 30,
+    "í": 26,
+    "ñ": 37,
+    "ó": 32,
+    "ú": 25,
+    "—": 27
+  },
+  "tng": {
+    "'": 27,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 20,
+    "d": 10,
+    "e": 12,
+    "g": 6,
+    "h": 26,
+    "i": 8,
+    "j": 24,
+    "k": 13,
+    "l": 16,
+    "m": 15,
+    "n": 7,
+    "o": 9,
+    "p": 25,
+    "r": 18,
+    "s": 19,
+    "t": 23,
+    "u": 11,
+    "v": 37,
+    "w": 21,
+    "y": 22,
+    "|": 4,
+    "â": 29,
+    "ç": 33,
+    "é": 14,
+    "ê": 31,
+    "ë": 34,
+    "î": 30,
+    "ô": 28,
+    "û": 32,
+    "ÿ": 35,
+    "ū": 38
+  },
+  "tnk": {
+    "'": 30,
+    "-": 26,
+    "0": 31,
+    "3": 35,
+    "4": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 27,
+    "d": 28,
+    "e": 11,
+    "f": 23,
+    "g": 32,
+    "h": 13,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 24,
+    "m": 9,
+    "n": 8,
+    "o": 20,
+    "p": 18,
+    "r": 10,
+    "s": 17,
+    "t": 16,
+    "u": 14,
+    "v": 21,
+    "w": 19,
+    "y": 29,
+    "|": 4,
+    "ǝ": 33,
+    "ə": 5,
+    "ɡ": 22,
+    "ɨ": 15
+  },
+  "tnn": {
+    "'": 28,
+    "-": 22,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 26,
+    "d": 27,
+    "e": 14,
+    "f": 24,
+    "g": 21,
+    "h": 15,
+    "i": 10,
+    "j": 25,
+    "k": 11,
+    "l": 17,
+    "m": 9,
+    "n": 6,
+    "o": 12,
+    "p": 20,
+    "r": 18,
+    "s": 19,
+    "t": 7,
+    "u": 16,
+    "v": 29,
+    "w": 23,
+    "|": 4,
+    "ə": 5,
+    "ɨ": 13
+  },
+  "tnp": {
+    "'": 28,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 27,
+    "d": 26,
+    "e": 17,
+    "f": 23,
+    "g": 21,
+    "h": 13,
+    "i": 7,
+    "j": 25,
+    "k": 12,
+    "l": 16,
+    "m": 8,
+    "n": 6,
+    "o": 15,
+    "p": 20,
+    "r": 18,
+    "s": 19,
+    "t": 9,
+    "u": 11,
+    "v": 22,
+    "w": 29,
+    "|": 4,
+    "ə": 5,
+    "ɨ": 14,
+    "—": 30
+  },
+  "tnr": {
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 30,
+    "d": 28,
+    "e": 8,
+    "f": 32,
+    "g": 12,
+    "h": 29,
+    "i": 16,
+    "j": 35,
+    "k": 14,
+    "l": 13,
+    "m": 9,
+    "n": 6,
+    "o": 10,
+    "p": 31,
+    "r": 11,
+    "s": 24,
+    "t": 26,
+    "u": 19,
+    "w": 25,
+    "y": 22,
+    "|": 4,
+    "é": 20,
+    "ë": 7,
+    "ñ": 33,
+    "ó": 23,
+    "ü": 17,
+    "ŋ": 18,
+    "ŝ": 27,
+    "ƴ": 36,
+    "ɓ": 15,
+    "ɗ": 21,
+    "–": 39,
+    "—": 37
+  },
+  "tnt": {
+    "'": 18,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 25,
+    "c": 22,
+    "d": 26,
+    "e": 15,
+    "f": 29,
+    "g": 20,
+    "h": 27,
+    "i": 6,
+    "j": 28,
+    "k": 13,
+    "l": 21,
+    "m": 10,
+    "n": 7,
+    "o": 11,
+    "p": 19,
+    "r": 16,
+    "s": 9,
+    "t": 12,
+    "u": 8,
+    "w": 17,
+    "y": 23,
+    "z": 30,
+    "|": 4,
+    "é": 14,
+    "ó": 31
+  },
+  "tob": {
+    "'": 8,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 30,
+    "c": 12,
+    "d": 20,
+    "e": 6,
+    "f": 34,
+    "g": 23,
+    "h": 18,
+    "i": 13,
+    "j": 27,
+    "k": 39,
+    "l": 16,
+    "m": 17,
+    "n": 9,
+    "o": 7,
+    "p": 24,
+    "q": 10,
+    "r": 26,
+    "s": 15,
+    "t": 11,
+    "u": 14,
+    "v": 28,
+    "x": 19,
+    "y": 22,
+    "z": 37,
+    "|": 5,
+    "á": 38,
+    "é": 32,
+    "í": 33,
+    "ñ": 25,
+    "ó": 36,
+    "ú": 29,
+    "ỹ": 21,
+    "–": 31
+  },
+  "toc": {
+    "'": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 31,
+    "c": 11,
+    "d": 29,
+    "e": 23,
+    "f": 35,
+    "g": 33,
+    "h": 13,
+    "i": 9,
+    "j": 27,
+    "k": 19,
+    "l": 12,
+    "m": 16,
+    "n": 7,
+    "o": 21,
+    "p": 17,
+    "q": 22,
+    "r": 24,
+    "s": 18,
+    "t": 10,
+    "u": 8,
+    "v": 36,
+    "w": 39,
+    "x": 15,
+    "y": 20,
+    "z": 28,
+    "|": 5,
+    "á": 14,
+    "é": 30,
+    "í": 25,
+    "ñ": 38,
+    "ó": 34,
+    "ú": 26,
+    "̲": 6,
+    "—": 32
+  },
+  "toh": {
+    "'": 28,
+    "-": 29,
+    "1": 33,
+    "5": 34,
+    "9": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "d": 16,
+    "e": 10,
+    "f": 26,
+    "g": 9,
+    "h": 11,
+    "i": 6,
+    "j": 27,
+    "k": 19,
+    "l": 13,
+    "m": 14,
+    "n": 8,
+    "o": 12,
+    "p": 25,
+    "r": 23,
+    "s": 18,
+    "t": 22,
+    "u": 7,
+    "v": 17,
+    "w": 20,
+    "x": 30,
+    "y": 15,
+    "z": 24,
+    "|": 4,
+    "–": 32
+  },
+  "tom": {
+    "'": 21,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 30,
+    "d": 26,
+    "e": 16,
+    "f": 28,
+    "g": 18,
+    "h": 22,
+    "i": 7,
+    "j": 29,
+    "k": 14,
+    "l": 19,
+    "m": 9,
+    "n": 6,
+    "o": 11,
+    "p": 17,
+    "r": 20,
+    "s": 13,
+    "t": 12,
+    "u": 8,
+    "w": 15,
+    "y": 25,
+    "z": 23,
+    "|": 4,
+    "é": 10
+  },
+  "tos": {
+    "'": 37,
+    "-": 28,
+    "0": 39,
+    "1": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 32,
+    "c": 11,
+    "d": 29,
+    "e": 27,
+    "f": 35,
+    "g": 33,
+    "h": 12,
+    "i": 7,
+    "j": 24,
+    "k": 16,
+    "l": 13,
+    "m": 15,
+    "n": 9,
+    "o": 21,
+    "p": 17,
+    "q": 22,
+    "r": 30,
+    "s": 19,
+    "t": 10,
+    "u": 8,
+    "v": 36,
+    "x": 14,
+    "y": 20,
+    "z": 26,
+    "|": 6,
+    "á": 18,
+    "é": 34,
+    "ì": 41,
+    "í": 23,
+    "ï": 40,
+    "ñ": 38,
+    "ó": 31,
+    "ú": 25,
+    "̱": 5
+  },
+  "tpi": {
+    "'": 36,
+    "-": 28,
+    "0": 27,
+    "1": 29,
+    "2": 30,
+    "3": 33,
+    "4": 32,
+    "5": 34,
+    "6": 35,
+    "7": 31,
+    "8": 38,
+    "9": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "d": 20,
+    "e": 11,
+    "f": 26,
+    "g": 13,
+    "h": 23,
+    "i": 5,
+    "j": 25,
+    "k": 16,
+    "l": 8,
+    "m": 10,
+    "n": 7,
+    "o": 9,
+    "p": 15,
+    "q": 39,
+    "r": 19,
+    "s": 12,
+    "t": 14,
+    "u": 18,
+    "v": 24,
+    "w": 22,
+    "y": 21,
+    "|": 4
+  },
+  "tpm": {
+    "'": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 26,
+    "d": 21,
+    "e": 18,
+    "f": 24,
+    "g": 25,
+    "h": 19,
+    "i": 6,
+    "j": 31,
+    "k": 11,
+    "l": 17,
+    "m": 9,
+    "n": 7,
+    "o": 10,
+    "p": 28,
+    "r": 23,
+    "s": 13,
+    "t": 14,
+    "u": 8,
+    "v": 30,
+    "w": 20,
+    "y": 16,
+    "z": 29,
+    "|": 4,
+    "ŋ": 27,
+    "ɔ": 22,
+    "ɛ": 12
+  },
+  "tpp": {
+    "'": 10,
+    "-": 46,
+    "0": 38,
+    "1": 41,
+    "2": 45,
+    "3": 47,
+    "4": 39,
+    "6": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 44,
+    "a": 4,
+    "b": 26,
+    "c": 14,
+    "d": 25,
+    "e": 23,
+    "f": 31,
+    "g": 29,
+    "h": 9,
+    "i": 6,
+    "j": 18,
+    "k": 13,
+    "l": 11,
+    "m": 19,
+    "n": 7,
+    "o": 22,
+    "p": 20,
+    "q": 37,
+    "r": 24,
+    "s": 16,
+    "t": 8,
+    "u": 12,
+    "v": 34,
+    "w": 21,
+    "x": 17,
+    "y": 15,
+    "z": 35,
+    "|": 5,
+    "á": 36,
+    "é": 30,
+    "í": 33,
+    "ñ": 43,
+    "ó": 32,
+    "ú": 27,
+    "́": 40,
+    "—": 28
+  },
+  "tpt": {
+    "-": 43,
+    "0": 37,
+    "1": 39,
+    "2": 42,
+    "3": 40,
+    "4": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "c": 15,
+    "d": 26,
+    "e": 24,
+    "f": 33,
+    "g": 32,
+    "h": 13,
+    "i": 7,
+    "j": 18,
+    "k": 10,
+    "l": 11,
+    "m": 21,
+    "n": 6,
+    "o": 19,
+    "p": 20,
+    "q": 22,
+    "r": 25,
+    "s": 16,
+    "t": 8,
+    "u": 9,
+    "v": 23,
+    "x": 17,
+    "y": 14,
+    "z": 36,
+    "|": 5,
+    "á": 35,
+    "é": 31,
+    "í": 30,
+    "ñ": 41,
+    "ó": 34,
+    "ú": 27,
+    "ʼ": 12,
+    "—": 29
+  },
+  "trc": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 15,
+    "d": 21,
+    "e": 11,
+    "f": 37,
+    "g": 25,
+    "h": 18,
+    "i": 9,
+    "j": 10,
+    "k": 42,
+    "l": 31,
+    "m": 17,
+    "n": 6,
+    "o": 14,
+    "p": 33,
+    "q": 28,
+    "r": 16,
+    "s": 12,
+    "t": 19,
+    "u": 13,
+    "v": 24,
+    "x": 29,
+    "y": 22,
+    "z": 32,
+    "|": 4,
+    "á": 23,
+    "â": 41,
+    "é": 27,
+    "ê": 40,
+    "í": 20,
+    "ñ": 39,
+    "ó": 26,
+    "ô": 43,
+    "ú": 30,
+    "û": 38,
+    "ü": 36,
+    "̱": 7,
+    "‐": 35,
+    "ꞌ": 8
+  },
+  "tri": {
+    "'": 22,
+    "-": 24,
+    "0": 25,
+    "1": 26,
+    "2": 27,
+    "3": 28,
+    "4": 30,
+    "5": 31,
+    "6": 32,
+    "7": 29,
+    "8": 34,
+    "9": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "e": 7,
+    "f": 39,
+    "g": 38,
+    "h": 17,
+    "i": 11,
+    "j": 18,
+    "k": 13,
+    "l": 35,
+    "m": 14,
+    "n": 6,
+    "o": 9,
+    "p": 15,
+    "r": 12,
+    "s": 21,
+    "t": 10,
+    "u": 20,
+    "w": 19,
+    "y": 37,
+    "z": 36,
+    "|": 4,
+    "ë": 8,
+    "ï": 16,
+    "ƒ": 40,
+    "ʻ": 23
+  },
+  "trn": {
+    "'": 15,
+    "1": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 33,
+    "c": 11,
+    "d": 32,
+    "e": 7,
+    "f": 35,
+    "g": 22,
+    "h": 19,
+    "i": 8,
+    "j": 18,
+    "k": 39,
+    "l": 31,
+    "m": 12,
+    "n": 10,
+    "o": 5,
+    "p": 16,
+    "q": 23,
+    "r": 13,
+    "s": 20,
+    "t": 9,
+    "u": 14,
+    "v": 17,
+    "x": 37,
+    "y": 21,
+    "z": 36,
+    "|": 4,
+    "á": 30,
+    "é": 24,
+    "í": 29,
+    "ñ": 25,
+    "ó": 28,
+    "ú": 26,
+    "‐": 34,
+    "—": 27
+  },
+  "trs": {
+    "'": 9,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 43,
+    "c": 30,
+    "d": 17,
+    "e": 15,
+    "f": 50,
+    "g": 12,
+    "h": 14,
+    "i": 7,
+    "j": 8,
+    "k": 23,
+    "l": 36,
+    "m": 19,
+    "n": 5,
+    "o": 24,
+    "p": 42,
+    "q": 54,
+    "r": 16,
+    "s": 13,
+    "t": 22,
+    "u": 10,
+    "v": 53,
+    "x": 49,
+    "y": 26,
+    "z": 52,
+    "|": 4,
+    "à": 18,
+    "á": 35,
+    "â": 21,
+    "è": 37,
+    "é": 47,
+    "ê": 33,
+    "ë": 51,
+    "ì": 25,
+    "í": 31,
+    "î": 34,
+    "ï": 28,
+    "ñ": 20,
+    "ò": 46,
+    "ó": 44,
+    "ô": 27,
+    "ù": 32,
+    "ú": 38,
+    "û": 29,
+    "̀": 41,
+    "́": 55,
+    "̂": 39,
+    "̱": 11,
+    "ḯ": 48,
+    "—": 45
+  },
+  "tso": {
+    "'": 28,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 30,
+    "d": 24,
+    "e": 9,
+    "f": 25,
+    "g": 21,
+    "h": 12,
+    "i": 6,
+    "j": 29,
+    "k": 8,
+    "l": 11,
+    "m": 14,
+    "n": 7,
+    "o": 15,
+    "p": 27,
+    "q": 32,
+    "r": 20,
+    "s": 16,
+    "t": 17,
+    "u": 10,
+    "v": 13,
+    "w": 18,
+    "x": 23,
+    "y": 19,
+    "z": 26,
+    "|": 5
+  },
+  "tsz": {
+    "'": 29,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 38,
+    "a": 4,
+    "b": 24,
+    "c": 19,
+    "d": 21,
+    "e": 9,
+    "f": 33,
+    "g": 22,
+    "h": 14,
+    "i": 6,
+    "j": 15,
+    "k": 8,
+    "l": 31,
+    "m": 16,
+    "n": 7,
+    "o": 23,
+    "p": 17,
+    "r": 12,
+    "s": 11,
+    "t": 13,
+    "u": 10,
+    "v": 37,
+    "x": 28,
+    "|": 5,
+    "á": 20,
+    "é": 25,
+    "í": 26,
+    "î": 34,
+    "ï": 18,
+    "ó": 27,
+    "ú": 30,
+    "́": 36,
+    "–": 32
+  },
+  "ttc": {
+    "'": 7,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 26,
+    "d": 28,
+    "e": 10,
+    "f": 33,
+    "g": 29,
+    "h": 25,
+    "i": 8,
+    "j": 12,
+    "k": 11,
+    "l": 14,
+    "m": 23,
+    "n": 9,
+    "o": 13,
+    "p": 24,
+    "q": 18,
+    "r": 27,
+    "s": 21,
+    "t": 5,
+    "u": 17,
+    "v": 39,
+    "w": 22,
+    "x": 16,
+    "y": 15,
+    "z": 19,
+    "|": 4,
+    "á": 31,
+    "é": 32,
+    "í": 35,
+    "ñ": 37,
+    "ó": 34,
+    "ú": 36,
+    "—": 38
+  },
+  "tte": {
+    "'": 24,
+    "-": 22,
+    "0": 27,
+    "1": 26,
+    "3": 28,
+    "4": 30,
+    "6": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "d": 20,
+    "e": 10,
+    "f": 29,
+    "g": 19,
+    "i": 6,
+    "k": 11,
+    "l": 9,
+    "m": 15,
+    "n": 7,
+    "o": 8,
+    "p": 21,
+    "s": 16,
+    "t": 18,
+    "u": 13,
+    "v": 23,
+    "w": 14,
+    "y": 12,
+    "|": 5
+  },
+  "ttq-script_tifinagh": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ⴰ": 13,
+    "ⴶ": 17,
+    "ⴹ": 7,
+    "ⴼ": 20,
+    "ⴾ": 16,
+    "ⵀ": 22,
+    "ⵂ": 23,
+    "ⵆ": 24,
+    "ⵈ": 25,
+    "ⵌ": 26,
+    "ⵍ": 10,
+    "ⵎ": 11,
+    "ⵏ": 5,
+    "ⵑ": 28,
+    "ⵓ": 9,
+    "ⵔ": 12,
+    "ⵗ": 15,
+    "ⵙ": 8,
+    "ⵛ": 21,
+    "ⵜ": 6,
+    "ⵢ": 14,
+    "ⵤ": 18,
+    "⵿": 19
+  },
+  "tue": {
+    "'": 46,
+    "-": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 26,
+    "c": 11,
+    "d": 21,
+    "e": 9,
+    "f": 42,
+    "g": 22,
+    "h": 39,
+    "i": 6,
+    "j": 24,
+    "k": 48,
+    "l": 38,
+    "m": 15,
+    "n": 17,
+    "o": 14,
+    "p": 19,
+    "q": 37,
+    "r": 5,
+    "s": 20,
+    "t": 13,
+    "u": 10,
+    "v": 43,
+    "w": 25,
+    "x": 47,
+    "y": 23,
+    "z": 44,
+    "|": 4,
+    "á": 31,
+    "ã": 12,
+    "é": 30,
+    "í": 33,
+    "ñ": 28,
+    "ó": 34,
+    "õ": 27,
+    "ú": 32,
+    "ĩ": 16,
+    "ũ": 18,
+    "́": 35,
+    "̶": 8,
+    "ṍ": 40,
+    "ṹ": 36,
+    "ẽ": 29,
+    "—": 41
+  },
+  "tuf": {
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 6,
+    "d": 28,
+    "e": 14,
+    "f": 34,
+    "g": 33,
+    "h": 15,
+    "i": 9,
+    "j": 17,
+    "k": 38,
+    "l": 27,
+    "m": 24,
+    "n": 13,
+    "o": 12,
+    "p": 29,
+    "q": 22,
+    "r": 7,
+    "s": 16,
+    "t": 8,
+    "u": 10,
+    "v": 35,
+    "w": 19,
+    "x": 37,
+    "y": 11,
+    "z": 36,
+    "|": 4,
+    "á": 18,
+    "é": 25,
+    "í": 21,
+    "ñ": 39,
+    "ó": 30,
+    "ú": 26,
+    "̃": 23,
+    "—": 31
+  },
+  "tuk-script_arabic": {
+    "'": 42,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "آ": 23,
+    "ئ": 37,
+    "ا": 6,
+    "ب": 15,
+    "ت": 17,
+    "ث": 38,
+    "ج": 28,
+    "ح": 29,
+    "خ": 27,
+    "د": 10,
+    "ذ": 40,
+    "ر": 8,
+    "ز": 19,
+    "س": 14,
+    "ش": 20,
+    "ص": 33,
+    "ض": 35,
+    "ط": 32,
+    "ظ": 36,
+    "ع": 26,
+    "غ": 24,
+    "ف": 31,
+    "ق": 18,
+    "ل": 9,
+    "م": 13,
+    "ن": 7,
+    "ه": 16,
+    "و": 11,
+    "ٰ": 39,
+    "پ": 21,
+    "چ": 25,
+    "ژ": 43,
+    "ک": 22,
+    "گ": 12,
+    "ی": 4,
+    "‍": 34,
+    "–": 41
+  },
+  "tuk-script_latin": {
+    "'": 38,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "d": 12,
+    "e": 6,
+    "f": 34,
+    "g": 17,
+    "h": 24,
+    "i": 7,
+    "j": 33,
+    "k": 20,
+    "l": 9,
+    "m": 14,
+    "n": 10,
+    "o": 18,
+    "p": 28,
+    "r": 8,
+    "s": 16,
+    "t": 22,
+    "u": 21,
+    "w": 31,
+    "y": 11,
+    "z": 23,
+    "|": 4,
+    "ä": 26,
+    "ç": 30,
+    "ö": 29,
+    "ü": 27,
+    "ý": 13,
+    "ň": 19,
+    "ş": 25,
+    "ž": 35,
+    "‐": 37,
+    "–": 36
+  },
+  "tuo": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 28,
+    "c": 13,
+    "d": 29,
+    "e": 6,
+    "f": 42,
+    "g": 27,
+    "h": 43,
+    "i": 8,
+    "j": 20,
+    "k": 48,
+    "l": 35,
+    "m": 17,
+    "n": 15,
+    "o": 11,
+    "p": 21,
+    "q": 31,
+    "r": 7,
+    "s": 16,
+    "t": 14,
+    "u": 19,
+    "v": 44,
+    "w": 18,
+    "x": 49,
+    "y": 26,
+    "z": 46,
+    "|": 4,
+    "á": 41,
+    "ã": 12,
+    "é": 37,
+    "í": 33,
+    "ñ": 30,
+    "ó": 40,
+    "õ": 25,
+    "ú": 36,
+    "ĩ": 22,
+    "ũ": 32,
+    "ʉ": 9,
+    "́": 34,
+    "̃": 23,
+    "̶": 39,
+    "ṍ": 45,
+    "ṹ": 47,
+    "ẽ": 24,
+    "—": 38
+  },
+  "tur": {
+    "!": 69,
+    "\"": 58,
+    "%": 56,
+    "'": 33,
+    "+": 71,
+    ",": 51,
+    "-": 48,
+    ".": 39,
+    "/": 50,
+    "0": 34,
+    "1": 35,
+    "2": 38,
+    "3": 44,
+    "4": 43,
+    "5": 42,
+    "6": 45,
+    "7": 47,
+    "8": 46,
+    "9": 40,
+    ":": 52,
+    ";": 54,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 70,
+    "a": 5,
+    "b": 20,
+    "c": 29,
+    "d": 13,
+    "e": 6,
+    "f": 32,
+    "g": 23,
+    "h": 27,
+    "i": 7,
+    "j": 36,
+    "k": 12,
+    "l": 10,
+    "m": 17,
+    "n": 8,
+    "o": 19,
+    "p": 30,
+    "q": 53,
+    "r": 9,
+    "s": 16,
+    "t": 14,
+    "u": 18,
+    "v": 25,
+    "w": 41,
+    "x": 49,
+    "y": 15,
+    "z": 24,
+    "|": 4,
+    "¥": 60,
+    "°": 66,
+    "á": 75,
+    "â": 55,
+    "ã": 64,
+    "ç": 26,
+    "é": 65,
+    "í": 59,
+    "î": 73,
+    "ó": 63,
+    "ö": 31,
+    "ú": 67,
+    "û": 74,
+    "ü": 21,
+    "ğ": 28,
+    "ı": 11,
+    "ł": 72,
+    "ş": 22,
+    "̇": 37,
+    "–": 62,
+    "—": 68,
+    "’": 57,
+    "”": 61
+  },
+  "tvw": {
+    "'": 27,
+    "-": 22,
+    "0": 31,
+    "4": 29,
+    "5": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 28,
+    "d": 20,
+    "e": 10,
+    "f": 26,
+    "g": 19,
+    "h": 24,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 17,
+    "m": 11,
+    "n": 8,
+    "o": 9,
+    "p": 13,
+    "r": 15,
+    "s": 16,
+    "t": 14,
+    "u": 7,
+    "w": 21,
+    "y": 23,
+    "|": 5
+  },
+  "twb": {
+    "'": 25,
+    "-": 32,
+    "0": 26,
+    "1": 28,
+    "2": 30,
+    "3": 31,
+    "4": 27,
+    "6": 29,
+    "7": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "d": 17,
+    "e": 18,
+    "f": 13,
+    "g": 8,
+    "i": 12,
+    "k": 16,
+    "l": 15,
+    "m": 14,
+    "n": 6,
+    "o": 21,
+    "p": 23,
+    "r": 22,
+    "s": 11,
+    "t": 9,
+    "u": 7,
+    "v": 24,
+    "w": 20,
+    "y": 10,
+    "|": 5
+  },
+  "twe": {
+    "'": 13,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 31,
+    "d": 23,
+    "e": 18,
+    "f": 27,
+    "g": 8,
+    "h": 22,
+    "i": 7,
+    "j": 29,
+    "k": 24,
+    "l": 15,
+    "m": 12,
+    "n": 6,
+    "o": 17,
+    "p": 25,
+    "q": 26,
+    "r": 10,
+    "s": 11,
+    "t": 14,
+    "u": 9,
+    "v": 30,
+    "w": 16,
+    "x": 19,
+    "y": 20,
+    "|": 5
+  },
+  "twu": {
+    "'": 19,
+    "-": 21,
+    "9": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 29,
+    "d": 16,
+    "e": 7,
+    "f": 20,
+    "g": 22,
+    "h": 17,
+    "i": 10,
+    "j": 28,
+    "k": 15,
+    "l": 9,
+    "m": 12,
+    "n": 6,
+    "o": 8,
+    "p": 23,
+    "r": 25,
+    "s": 13,
+    "t": 14,
+    "u": 11,
+    "w": 27,
+    "y": 24,
+    "z": 26,
+    "|": 5
+  },
+  "txa": {
+    "'": 22,
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 21,
+    "d": 17,
+    "g": 13,
+    "h": 25,
+    "i": 8,
+    "j": 24,
+    "k": 11,
+    "l": 19,
+    "m": 12,
+    "n": 6,
+    "o": 5,
+    "p": 18,
+    "r": 14,
+    "s": 10,
+    "t": 15,
+    "u": 9,
+    "w": 20,
+    "y": 16,
+    "|": 4
+  },
+  "txq": {
+    "'": 27,
+    "-": 22,
+    "0": 28,
+    "1": 33,
+    "2": 31,
+    "3": 38,
+    "4": 32,
+    "5": 36,
+    "6": 34,
+    "7": 37,
+    "8": 40,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 29,
+    "d": 18,
+    "e": 7,
+    "f": 21,
+    "g": 20,
+    "h": 17,
+    "i": 10,
+    "j": 30,
+    "k": 13,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 8,
+    "p": 24,
+    "r": 16,
+    "s": 15,
+    "t": 12,
+    "u": 9,
+    "w": 26,
+    "y": 25,
+    "z": 35,
+    "|": 5,
+    "—": 39,
+    "ꞌ": 23
+  },
+  "txu": {
+    "'": 29,
+    "0": 36,
+    "1": 37,
+    "2": 39,
+    "3": 42,
+    "4": 40,
+    "5": 41,
+    "6": 43,
+    "7": 38,
+    "8": 44,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "d": 18,
+    "e": 9,
+    "g": 28,
+    "h": 17,
+    "i": 16,
+    "j": 11,
+    "k": 8,
+    "m": 6,
+    "n": 7,
+    "o": 21,
+    "p": 23,
+    "r": 10,
+    "t": 12,
+    "u": 14,
+    "w": 20,
+    "x": 26,
+    "y": 25,
+    "|": 4,
+    "à": 30,
+    "ã": 13,
+    "ê": 24,
+    "ô": 31,
+    "õ": 32,
+    "ĩ": 27,
+    "ũ": 33,
+    "ẽ": 22,
+    "ỳ": 19,
+    "ỹ": 34,
+    "—": 35
+  },
+  "tye": {
+    "'": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 33,
+    "d": 13,
+    "e": 26,
+    "f": 29,
+    "g": 15,
+    "h": 31,
+    "i": 6,
+    "j": 34,
+    "k": 8,
+    "l": 16,
+    "m": 14,
+    "n": 7,
+    "o": 28,
+    "p": 36,
+    "r": 24,
+    "s": 21,
+    "t": 25,
+    "u": 10,
+    "v": 42,
+    "w": 23,
+    "y": 19,
+    "z": 22,
+    "|": 4,
+    "à": 17,
+    "á": 39,
+    "ã": 27,
+    "è": 43,
+    "é": 50,
+    "ì": 18,
+    "í": 46,
+    "ò": 37,
+    "ó": 44,
+    "ù": 38,
+    "ú": 48,
+    "ĩ": 40,
+    "ń": 45,
+    "ũ": 20,
+    "ǹ": 41,
+    "ɔ": 11,
+    "ɛ": 9,
+    "̀": 30,
+    "́": 47,
+    "̃": 35,
+    "ḿ": 49
+  },
+  "tzh-dialect_bachajon": {
+    "'": 11,
+    "-": 27,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 7,
+    "d": 26,
+    "e": 8,
+    "f": 33,
+    "g": 29,
+    "h": 13,
+    "i": 9,
+    "j": 20,
+    "l": 10,
+    "m": 19,
+    "n": 15,
+    "o": 14,
+    "p": 22,
+    "q": 25,
+    "r": 24,
+    "s": 16,
+    "t": 6,
+    "u": 12,
+    "v": 36,
+    "w": 23,
+    "x": 21,
+    "y": 17,
+    "z": 35,
+    "|": 4,
+    "á": 34,
+    "é": 31,
+    "í": 30,
+    "ñ": 37,
+    "ó": 32,
+    "ú": 28
+  },
+  "tzh-dialect_tenejapa": {
+    "'": 11,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 9,
+    "d": 26,
+    "e": 6,
+    "f": 32,
+    "g": 29,
+    "h": 21,
+    "i": 8,
+    "j": 17,
+    "k": 39,
+    "l": 10,
+    "m": 19,
+    "n": 16,
+    "o": 13,
+    "p": 23,
+    "q": 22,
+    "r": 25,
+    "s": 15,
+    "t": 7,
+    "u": 12,
+    "v": 36,
+    "w": 24,
+    "x": 20,
+    "y": 14,
+    "z": 35,
+    "|": 4,
+    "á": 38,
+    "é": 33,
+    "í": 34,
+    "ñ": 37,
+    "ó": 31,
+    "ú": 30,
+    "—": 27
+  },
+  "tzj-dialect_eastern": {
+    "'": 9,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 6,
+    "d": 28,
+    "e": 7,
+    "f": 36,
+    "g": 30,
+    "h": 17,
+    "i": 11,
+    "j": 10,
+    "k": 18,
+    "l": 16,
+    "m": 20,
+    "n": 8,
+    "o": 15,
+    "p": 26,
+    "q": 24,
+    "r": 12,
+    "s": 23,
+    "t": 13,
+    "u": 14,
+    "v": 38,
+    "w": 21,
+    "x": 19,
+    "y": 25,
+    "z": 27,
+    "|": 4,
+    "á": 32,
+    "é": 37,
+    "í": 34,
+    "ñ": 39,
+    "ó": 35,
+    "ú": 33,
+    "—": 31
+  },
+  "tzj-dialect_western": {
+    "'": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 9,
+    "d": 28,
+    "e": 11,
+    "f": 34,
+    "g": 29,
+    "h": 16,
+    "i": 6,
+    "j": 7,
+    "k": 18,
+    "l": 17,
+    "m": 19,
+    "n": 8,
+    "o": 12,
+    "p": 25,
+    "q": 23,
+    "r": 13,
+    "s": 24,
+    "t": 14,
+    "u": 15,
+    "v": 37,
+    "w": 21,
+    "x": 20,
+    "y": 27,
+    "z": 26,
+    "|": 4,
+    "á": 33,
+    "é": 36,
+    "í": 31,
+    "ñ": 39,
+    "ó": 35,
+    "ú": 32,
+    "̱": 38,
+    "—": 30
+  },
+  "tzo-dialect_chamula": {
+    "'": 11,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 8,
+    "d": 26,
+    "e": 12,
+    "f": 31,
+    "g": 30,
+    "h": 20,
+    "i": 7,
+    "j": 14,
+    "k": 37,
+    "l": 9,
+    "m": 21,
+    "n": 15,
+    "o": 13,
+    "p": 24,
+    "q": 23,
+    "r": 25,
+    "s": 16,
+    "t": 6,
+    "u": 10,
+    "v": 22,
+    "x": 17,
+    "y": 19,
+    "z": 34,
+    "|": 4,
+    "á": 32,
+    "é": 35,
+    "í": 29,
+    "ñ": 38,
+    "ó": 36,
+    "ú": 33,
+    "—": 27
+  },
+  "tzo-dialect_chenalho": {
+    "'": 10,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 6,
+    "d": 26,
+    "e": 13,
+    "f": 32,
+    "g": 28,
+    "h": 17,
+    "i": 7,
+    "j": 16,
+    "k": 37,
+    "l": 8,
+    "m": 21,
+    "n": 14,
+    "o": 12,
+    "p": 24,
+    "q": 23,
+    "r": 25,
+    "s": 15,
+    "t": 9,
+    "u": 11,
+    "v": 22,
+    "x": 20,
+    "y": 19,
+    "z": 29,
+    "|": 4,
+    "á": 36,
+    "é": 34,
+    "í": 31,
+    "ñ": 38,
+    "ó": 33,
+    "ú": 35,
+    "—": 27
+  },
+  "ubl": {
+    "'": 23,
+    "-": 25,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "3": 39,
+    "4": 36,
+    "5": 38,
+    "6": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 27,
+    "d": 15,
+    "e": 21,
+    "f": 28,
+    "g": 9,
+    "h": 26,
+    "i": 7,
+    "j": 24,
+    "k": 11,
+    "l": 20,
+    "m": 12,
+    "n": 6,
+    "o": 8,
+    "p": 16,
+    "q": 32,
+    "r": 17,
+    "s": 10,
+    "t": 14,
+    "u": 18,
+    "v": 30,
+    "w": 22,
+    "x": 35,
+    "y": 13,
+    "z": 29,
+    "|": 5
+  },
+  "ubu": {
+    "'": 30,
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "d": 20,
+    "e": 11,
+    "g": 15,
+    "i": 8,
+    "j": 24,
+    "k": 13,
+    "l": 12,
+    "m": 10,
+    "n": 7,
+    "o": 6,
+    "p": 14,
+    "r": 19,
+    "s": 16,
+    "t": 22,
+    "u": 9,
+    "w": 23,
+    "y": 17,
+    "|": 4,
+    "á": 25,
+    "é": 26,
+    "í": 29,
+    "ó": 28,
+    "ú": 27
+  },
+  "udm": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 9,
+    "б": 30,
+    "в": 22,
+    "г": 28,
+    "д": 21,
+    "е": 15,
+    "ж": 35,
+    "з": 17,
+    "и": 16,
+    "й": 32,
+    "к": 11,
+    "л": 10,
+    "м": 14,
+    "н": 8,
+    "о": 7,
+    "п": 24,
+    "р": 20,
+    "с": 5,
+    "т": 12,
+    "у": 13,
+    "ф": 41,
+    "х": 38,
+    "ц": 44,
+    "ч": 33,
+    "ш": 25,
+    "щ": 43,
+    "ъ": 29,
+    "ы": 6,
+    "ь": 18,
+    "э": 19,
+    "ю": 37,
+    "я": 27,
+    "ё": 26,
+    "ӝ": 42,
+    "ӟ": 36,
+    "ӥ": 23,
+    "ӧ": 31,
+    "ӵ": 39,
+    "–": 40
+  },
+  "udu": {
+    "'": 16,
+    "-": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 21,
+    "c": 27,
+    "d": 12,
+    "e": 14,
+    "f": 35,
+    "g": 20,
+    "h": 13,
+    "i": 5,
+    "j": 25,
+    "k": 10,
+    "l": 18,
+    "m": 7,
+    "n": 9,
+    "o": 8,
+    "p": 24,
+    "r": 23,
+    "s": 15,
+    "t": 17,
+    "u": 11,
+    "w": 19,
+    "y": 22,
+    "|": 4,
+    "ŋ": 30,
+    "̧": 34,
+    "̱": 26,
+    "ḵ": 28,
+    "ṯ": 29,
+    "ẖ": 31,
+    "–": 33,
+    "—": 32
+  },
+  "uig-script_arabic": {
+    "-": 36,
+    "0": 40,
+    "1": 41,
+    "2": 42,
+    "4": 45,
+    "5": 44,
+    "6": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ئ": 13,
+    "ا": 6,
+    "ب": 19,
+    "ت": 11,
+    "ج": 35,
+    "خ": 34,
+    "د": 16,
+    "ر": 10,
+    "ز": 27,
+    "س": 20,
+    "ش": 23,
+    "غ": 26,
+    "ف": 37,
+    "ق": 14,
+    "ك": 17,
+    "ل": 9,
+    "م": 15,
+    "ن": 8,
+    "و": 22,
+    "ى": 5,
+    "ي": 18,
+    "پ": 21,
+    "چ": 29,
+    "ژ": 39,
+    "ڭ": 25,
+    "گ": 30,
+    "ھ": 32,
+    "ۆ": 31,
+    "ۇ": 12,
+    "ۈ": 28,
+    "ۋ": 33,
+    "ې": 24,
+    "ە": 7,
+    "—": 38
+  },
+  "uig-script_cyrillic": {
+    "-": 37,
+    "0": 43,
+    "1": 44,
+    "2": 45,
+    "4": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "­": 35,
+    "а": 6,
+    "б": 17,
+    "в": 31,
+    "г": 30,
+    "д": 12,
+    "е": 24,
+    "ж": 39,
+    "з": 25,
+    "и": 5,
+    "й": 19,
+    "к": 18,
+    "л": 9,
+    "м": 14,
+    "н": 8,
+    "о": 27,
+    "п": 23,
+    "р": 10,
+    "с": 16,
+    "т": 13,
+    "у": 11,
+    "ф": 41,
+    "х": 32,
+    "ц": 50,
+    "ч": 29,
+    "ш": 20,
+    "ъ": 42,
+    "ы": 48,
+    "ь": 46,
+    "ю": 40,
+    "я": 34,
+    "ё": 49,
+    "ғ": 22,
+    "җ": 36,
+    "қ": 15,
+    "ң": 21,
+    "ү": 26,
+    "һ": 28,
+    "ә": 7,
+    "ө": 33,
+    "—": 38
+  },
+  "ukr": {
+    "!": 94,
+    "\"": 84,
+    "%": 77,
+    "&": 93,
+    "'": 40,
+    "+": 85,
+    ",": 65,
+    "-": 38,
+    ".": 49,
+    "/": 59,
+    "0": 37,
+    "1": 39,
+    "2": 41,
+    "3": 46,
+    "4": 44,
+    "5": 43,
+    "6": 45,
+    "7": 48,
+    "8": 47,
+    "9": 42,
+    ":": 58,
+    ";": 82,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 89,
+    "]": 90,
+    "a": 50,
+    "b": 66,
+    "c": 55,
+    "d": 72,
+    "e": 53,
+    "f": 76,
+    "g": 64,
+    "h": 62,
+    "i": 54,
+    "j": 83,
+    "k": 70,
+    "l": 63,
+    "m": 73,
+    "n": 51,
+    "o": 56,
+    "p": 57,
+    "q": 88,
+    "r": 52,
+    "s": 68,
+    "t": 60,
+    "u": 69,
+    "v": 71,
+    "w": 79,
+    "x": 80,
+    "y": 74,
+    "z": 86,
+    "|": 4,
+    "°": 92,
+    "²": 95,
+    "õ": 87,
+    "ú": 91,
+    "́": 96,
+    "а": 6,
+    "б": 24,
+    "в": 11,
+    "г": 25,
+    "д": 17,
+    "е": 13,
+    "ж": 30,
+    "з": 21,
+    "и": 9,
+    "й": 27,
+    "к": 15,
+    "л": 16,
+    "м": 19,
+    "н": 7,
+    "о": 5,
+    "п": 20,
+    "р": 12,
+    "с": 14,
+    "т": 10,
+    "у": 18,
+    "ф": 36,
+    "х": 28,
+    "ц": 31,
+    "ч": 26,
+    "ш": 32,
+    "щ": 35,
+    "ь": 23,
+    "ю": 29,
+    "я": 22,
+    "є": 34,
+    "і": 8,
+    "ї": 33,
+    "ґ": 61,
+    "​": 81,
+    "‐": 97,
+    "–": 75,
+    "—": 67,
+    "№": 78
+  },
+  "umb": {
+    "!": 69,
+    "$": 72,
+    "%": 65,
+    "&": 70,
+    "'": 24,
+    ",": 59,
+    "-": 31,
+    ".": 51,
+    "/": 58,
+    "0": 35,
+    "1": 36,
+    "2": 38,
+    "3": 43,
+    "4": 45,
+    "5": 44,
+    "6": 47,
+    "7": 52,
+    "8": 49,
+    "9": 40,
+    ":": 57,
+    ";": 66,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "?": 74,
+    "`": 68,
+    "a": 5,
+    "b": 23,
+    "c": 19,
+    "d": 20,
+    "e": 9,
+    "f": 28,
+    "g": 22,
+    "h": 25,
+    "i": 7,
+    "j": 27,
+    "k": 10,
+    "l": 8,
+    "m": 17,
+    "n": 11,
+    "o": 6,
+    "p": 21,
+    "q": 46,
+    "r": 26,
+    "s": 18,
+    "t": 15,
+    "u": 12,
+    "v": 13,
+    "w": 14,
+    "x": 50,
+    "y": 16,
+    "z": 39,
+    "|": 4,
+    "²": 71,
+    "º": 61,
+    "á": 42,
+    "â": 48,
+    "ã": 29,
+    "ç": 55,
+    "é": 41,
+    "ê": 63,
+    "í": 54,
+    "ñ": 33,
+    "ó": 53,
+    "ô": 73,
+    "õ": 32,
+    "ú": 60,
+    "ĩ": 30,
+    "ũ": 56,
+    "ẽ": 34,
+    "ῖ": 64,
+    "–": 67,
+    "’": 37,
+    "”": 62
+  },
+  "unr": {
+    "'": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ଁ": 33,
+    "ଂ": 56,
+    "ଃ": 11,
+    "ଅ": 34,
+    "ଆ": 14,
+    "ଇ": 19,
+    "ଉ": 38,
+    "ଏ": 21,
+    "କ": 8,
+    "ଖ": 43,
+    "ଗ": 20,
+    "ଘ": 53,
+    "ଙ": 29,
+    "ଚ": 31,
+    "ଛ": 52,
+    "ଜ": 28,
+    "ଝ": 55,
+    "ଞ": 48,
+    "ଟ": 37,
+    "ଠ": 46,
+    "ଡ": 22,
+    "ଢ": 54,
+    "ଣ": 40,
+    "ତ": 12,
+    "ଥ": 45,
+    "ଦ": 17,
+    "ଧ": 44,
+    "ନ": 9,
+    "ପ": 27,
+    "ଫ": 51,
+    "ବ": 24,
+    "ଭ": 47,
+    "ମ": 16,
+    "ଯ": 41,
+    "ର": 13,
+    "ଲ": 23,
+    "ଶ": 32,
+    "ଷ": 49,
+    "ସ": 26,
+    "ହ": 25,
+    "଼": 30,
+    "ା": 5,
+    "ି": 10,
+    "ୀ": 35,
+    "ୁ": 15,
+    "େ": 7,
+    "ୋ": 42,
+    "୍": 6,
+    "ୟ": 18,
+    "ୱ": 39,
+    "‍": 36
+  },
+  "upv": {
+    "'": 27,
+    "-": 25,
+    "0": 28,
+    "1": 29,
+    "2": 32,
+    "4": 30,
+    "5": 31,
+    "6": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 22,
+    "d": 24,
+    "e": 5,
+    "g": 16,
+    "i": 7,
+    "j": 20,
+    "k": 11,
+    "l": 13,
+    "m": 14,
+    "n": 6,
+    "o": 10,
+    "p": 18,
+    "r": 8,
+    "s": 12,
+    "t": 17,
+    "u": 15,
+    "v": 19,
+    "w": 21,
+    "|": 4,
+    "ö": 26,
+    "̃": 23
+  },
+  "ura": {
+    "-": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 9,
+    "d": 22,
+    "e": 7,
+    "f": 21,
+    "g": 26,
+    "h": 13,
+    "i": 8,
+    "j": 15,
+    "l": 17,
+    "m": 20,
+    "n": 6,
+    "o": 11,
+    "p": 24,
+    "q": 19,
+    "r": 12,
+    "s": 16,
+    "t": 14,
+    "u": 10,
+    "v": 27,
+    "y": 25,
+    "|": 5,
+    "ñ": 28
+  },
+  "urb": {
+    "'": 14,
+    "0": 36,
+    "1": 41,
+    "2": 40,
+    "3": 43,
+    "4": 39,
+    "5": 42,
+    "6": 44,
+    "7": 45,
+    "8": 47,
+    "9": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 33,
+    "d": 22,
+    "e": 6,
+    "g": 29,
+    "h": 9,
+    "i": 16,
+    "j": 15,
+    "k": 12,
+    "m": 10,
+    "n": 17,
+    "o": 21,
+    "p": 11,
+    "r": 13,
+    "s": 24,
+    "t": 8,
+    "u": 7,
+    "w": 23,
+    "x": 27,
+    "y": 18,
+    "|": 4,
+    "á": 31,
+    "ã": 20,
+    "é": 32,
+    "ó": 34,
+    "õ": 26,
+    "ú": 37,
+    "ý": 38,
+    "ĩ": 25,
+    "ũ": 28,
+    "ẽ": 19,
+    "ỹ": 35,
+    "—": 30
+  },
+  "urd-script_arabic": {
+    "!": 94,
+    "\"": 72,
+    "$": 103,
+    "%": 86,
+    "'": 68,
+    ",": 73,
+    "-": 62,
+    ".": 65,
+    "/": 67,
+    "0": 39,
+    "1": 42,
+    "2": 49,
+    "3": 54,
+    "4": 53,
+    "5": 51,
+    "6": 56,
+    "7": 60,
+    "8": 58,
+    "9": 50,
+    ":": 66,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 92,
+    "]": 93,
+    "a": 69,
+    "b": 90,
+    "c": 87,
+    "d": 83,
+    "e": 89,
+    "f": 98,
+    "g": 101,
+    "h": 106,
+    "i": 77,
+    "j": 105,
+    "k": 109,
+    "l": 84,
+    "m": 97,
+    "n": 76,
+    "o": 80,
+    "p": 88,
+    "r": 114,
+    "s": 75,
+    "u": 74,
+    "w": 99,
+    "x": 104,
+    "y": 79,
+    "|": 4,
+    "°": 96,
+    "õ": 108,
+    "،": 48,
+    "ؔ": 117,
+    "؛": 85,
+    "ء": 57,
+    "آ": 37,
+    "أ": 118,
+    "ؤ": 55,
+    "إ": 91,
+    "ئ": 24,
+    "ا": 5,
+    "ب": 17,
+    "ت": 14,
+    "ث": 41,
+    "ج": 21,
+    "ح": 31,
+    "خ": 34,
+    "د": 19,
+    "ذ": 45,
+    "ر": 8,
+    "ز": 29,
+    "س": 15,
+    "ش": 30,
+    "ص": 35,
+    "ض": 40,
+    "ط": 33,
+    "ظ": 43,
+    "ع": 27,
+    "غ": 44,
+    "ف": 28,
+    "ق": 26,
+    "ل": 16,
+    "م": 13,
+    "ن": 11,
+    "ه": 113,
+    "و": 9,
+    "ي": 111,
+    "ً": 61,
+    "ٍ": 95,
+    "َ": 63,
+    "ُ": 46,
+    "ِ": 52,
+    "ّ": 59,
+    "ٔ": 116,
+    "ٖ": 107,
+    "٪": 112,
+    "ٰ": 64,
+    "ٴ": 115,
+    "ٹ": 25,
+    "پ": 20,
+    "چ": 32,
+    "ڈ": 36,
+    "ڑ": 38,
+    "ژ": 71,
+    "ک": 7,
+    "گ": 22,
+    "ں": 18,
+    "ھ": 23,
+    "ہ": 12,
+    "ۂ": 82,
+    "ۃ": 81,
+    "ی": 6,
+    "ے": 10,
+    "۔": 47,
+    "۰": 119,
+    "۱": 126,
+    "۲": 123,
+    "۳": 125,
+    "۴": 121,
+    "۵": 124,
+    "۷": 122,
+    "​": 102,
+    "‏": 78,
+    "–": 120,
+    "—": 110,
+    "”": 70,
+    "ﷺ": 100
+  },
+  "urd-script_devanagari": {
+    "'": 50,
+    "-": 49,
+    "0": 60,
+    "1": 63,
+    "2": 62,
+    "3": 66,
+    "4": 64,
+    "5": 65,
+    "6": 70,
+    "7": 68,
+    "8": 69,
+    "9": 67,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 37,
+    "ं": 20,
+    "ः": 61,
+    "अ": 34,
+    "आ": 36,
+    "इ": 40,
+    "ई": 42,
+    "उ": 32,
+    "ऊ": 58,
+    "ए": 35,
+    "ऐ": 53,
+    "ओ": 55,
+    "औ": 33,
+    "क": 6,
+    "ख": 30,
+    "ग": 28,
+    "घ": 56,
+    "च": 41,
+    "छ": 47,
+    "ज": 22,
+    "झ": 48,
+    "ट": 51,
+    "ठ": 52,
+    "ड": 45,
+    "ढ": 59,
+    "ण": 71,
+    "त": 13,
+    "थ": 43,
+    "द": 21,
+    "ध": 57,
+    "न": 10,
+    "प": 26,
+    "फ": 39,
+    "ब": 24,
+    "भ": 44,
+    "म": 12,
+    "य": 25,
+    "र": 8,
+    "ल": 18,
+    "व": 29,
+    "श": 38,
+    "स": 11,
+    "ह": 9,
+    "़": 16,
+    "ा": 5,
+    "ि": 15,
+    "ी": 14,
+    "ु": 23,
+    "ू": 31,
+    "े": 7,
+    "ै": 27,
+    "ॉ": 73,
+    "ो": 17,
+    "ौ": 46,
+    "्": 19,
+    "‍": 72,
+    "—": 54
+  },
+  "urd-script_latin": {
+    "'": 39,
+    "-": 34,
+    "0": 40,
+    "1": 43,
+    "2": 42,
+    "3": 46,
+    "4": 41,
+    "5": 44,
+    "6": 49,
+    "7": 47,
+    "8": 48,
+    "9": 45,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 22,
+    "c": 30,
+    "d": 21,
+    "e": 8,
+    "f": 33,
+    "g": 26,
+    "h": 6,
+    "i": 10,
+    "j": 25,
+    "k": 9,
+    "l": 19,
+    "m": 16,
+    "n": 14,
+    "o": 20,
+    "p": 23,
+    "q": 31,
+    "r": 11,
+    "s": 12,
+    "t": 17,
+    "u": 13,
+    "w": 28,
+    "y": 24,
+    "z": 29,
+    "|": 4,
+    "ā": 7,
+    "ġ": 37,
+    "ī": 15,
+    "ū": 27,
+    "ḍ": 38,
+    "ḳ": 32,
+    "ṅ": 18,
+    "ṛ": 35,
+    "ṭ": 36
+  },
+  "urk": {
+    "'": 37,
+    "-": 33,
+    "0": 45,
+    "1": 48,
+    "2": 50,
+    "3": 44,
+    "4": 46,
+    "6": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 49,
+    "q": 51,
+    "t": 52,
+    "|": 4,
+    "ก": 10,
+    "ค": 36,
+    "ฆ": 28,
+    "ง": 30,
+    "จ": 24,
+    "ช": 43,
+    "ซ": 18,
+    "ฌ": 29,
+    "ญ": 23,
+    "ด": 14,
+    "ต": 16,
+    "ท": 41,
+    "น": 13,
+    "บ": 19,
+    "ป": 26,
+    "พ": 42,
+    "ฟ": 40,
+    "ม": 21,
+    "ย": 27,
+    "ร": 22,
+    "ล": 20,
+    "ว": 31,
+    "อ": 6,
+    "ฮ": 17,
+    "ะ": 8,
+    "ั": 11,
+    "า": 5,
+    "ิ": 32,
+    "ี": 9,
+    "ึ": 39,
+    "ื": 15,
+    "ุ": 38,
+    "ู": 7,
+    "เ": 12,
+    "แ": 34,
+    "โ": 25,
+    "ํ": 35
+  },
+  "urt": {
+    "'": 41,
+    "-": 35,
+    "0": 29,
+    "1": 31,
+    "2": 32,
+    "3": 36,
+    "4": 33,
+    "5": 34,
+    "6": 38,
+    "7": 37,
+    "8": 40,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 24,
+    "c": 27,
+    "d": 25,
+    "e": 5,
+    "f": 28,
+    "g": 18,
+    "h": 20,
+    "i": 6,
+    "j": 26,
+    "k": 15,
+    "l": 17,
+    "m": 10,
+    "n": 8,
+    "o": 12,
+    "p": 11,
+    "r": 22,
+    "s": 19,
+    "t": 13,
+    "u": 14,
+    "v": 30,
+    "w": 23,
+    "y": 9,
+    "|": 4,
+    "ŋ": 16,
+    "ꞌ": 21
+  },
+  "ury": {
+    "'": 31,
+    "-": 27,
+    "0": 34,
+    "1": 37,
+    "2": 36,
+    "3": 40,
+    "4": 35,
+    "5": 43,
+    "6": 39,
+    "7": 42,
+    "8": 44,
+    "9": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 33,
+    "d": 16,
+    "e": 7,
+    "f": 32,
+    "g": 14,
+    "h": 17,
+    "i": 9,
+    "j": 30,
+    "k": 8,
+    "l": 12,
+    "m": 10,
+    "n": 6,
+    "o": 13,
+    "p": 21,
+    "r": 22,
+    "s": 11,
+    "t": 20,
+    "u": 24,
+    "w": 15,
+    "y": 26,
+    "z": 18,
+    "|": 4,
+    "ë": 23,
+    "ï": 25,
+    "–": 28,
+    "—": 41,
+    "ꞌ": 29
+  },
+  "usp": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 10,
+    "d": 26,
+    "e": 15,
+    "f": 33,
+    "g": 31,
+    "h": 17,
+    "i": 6,
+    "j": 7,
+    "k": 9,
+    "l": 13,
+    "m": 22,
+    "n": 14,
+    "o": 16,
+    "p": 25,
+    "q": 28,
+    "r": 12,
+    "s": 18,
+    "t": 11,
+    "u": 19,
+    "v": 35,
+    "w": 20,
+    "x": 23,
+    "y": 24,
+    "z": 27,
+    "|": 4,
+    "á": 36,
+    "é": 30,
+    "í": 32,
+    "ñ": 37,
+    "ó": 34,
+    "ú": 29
+  },
+  "uzb-script_cyrillic": {
+    "-": 40,
+    "0": 41,
+    "1": 42,
+    "2": 43,
+    "3": 48,
+    "4": 44,
+    "5": 49,
+    "6": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 5,
+    "б": 13,
+    "в": 27,
+    "г": 12,
+    "д": 10,
+    "е": 23,
+    "ж": 34,
+    "з": 20,
+    "и": 6,
+    "й": 25,
+    "к": 19,
+    "л": 8,
+    "м": 15,
+    "н": 7,
+    "о": 11,
+    "п": 32,
+    "р": 9,
+    "с": 17,
+    "т": 16,
+    "у": 14,
+    "ф": 35,
+    "х": 28,
+    "ц": 47,
+    "ч": 26,
+    "ш": 22,
+    "ъ": 38,
+    "ь": 45,
+    "э": 29,
+    "ю": 37,
+    "я": 31,
+    "ё": 30,
+    "ў": 21,
+    "ғ": 33,
+    "қ": 18,
+    "ҳ": 24,
+    "–": 36,
+    "—": 39
+  },
+  "uzb-script_latin": {
+    "!": 61,
+    "\"": 46,
+    "%": 52,
+    "&": 64,
+    "'": 33,
+    "+": 57,
+    ",": 50,
+    "-": 31,
+    ".": 44,
+    "/": 45,
+    "0": 32,
+    "1": 35,
+    "2": 36,
+    "3": 40,
+    "4": 39,
+    "5": 38,
+    "6": 41,
+    "7": 43,
+    "8": 42,
+    "9": 37,
+    ":": 48,
+    ";": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 58,
+    "\\": 75,
+    "]": 59,
+    "`": 65,
+    "a": 4,
+    "b": 20,
+    "c": 26,
+    "d": 14,
+    "e": 21,
+    "f": 28,
+    "g": 15,
+    "h": 12,
+    "i": 6,
+    "j": 29,
+    "k": 19,
+    "l": 9,
+    "m": 18,
+    "n": 7,
+    "o": 8,
+    "p": 27,
+    "q": 22,
+    "r": 10,
+    "s": 11,
+    "t": 13,
+    "u": 16,
+    "v": 24,
+    "w": 47,
+    "x": 30,
+    "y": 17,
+    "z": 25,
+    "|": 5,
+    "": 62,
+    "¥": 55,
+    "°": 63,
+    "²": 56,
+    "ë": 110,
+    "ò": 94,
+    "ó": 90,
+    "õ": 54,
+    "ģ": 93,
+    "ı": 111,
+    "ş": 105,
+    "ƣ": 103,
+    "ƶ": 112,
+    "ʙ": 87,
+    "ʻ": 23,
+    "ʼ": 34,
+    "́": 68,
+    "μ": 104,
+    "а": 66,
+    "б": 80,
+    "в": 95,
+    "г": 77,
+    "д": 84,
+    "е": 86,
+    "з": 88,
+    "и": 67,
+    "й": 89,
+    "к": 79,
+    "л": 83,
+    "м": 71,
+    "н": 70,
+    "о": 73,
+    "п": 106,
+    "р": 69,
+    "с": 76,
+    "т": 72,
+    "у": 74,
+    "х": 60,
+    "ч": 85,
+    "ш": 98,
+    "ы": 108,
+    "ь": 96,
+    "э": 101,
+    "ё": 107,
+    "ј": 102,
+    "ў": 82,
+    "ғ": 91,
+    "қ": 78,
+    "ҳ": 81,
+    "ӯ": 97,
+    "‒": 92,
+    "–": 51,
+    "—": 49,
+    "♂": 109,
+    "✔": 99,
+    "️": 100
+  },
+  "vag": {
+    "'": 39,
+    "2": 43,
+    "3": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 29,
+    "d": 14,
+    "e": 17,
+    "f": 28,
+    "g": 19,
+    "h": 21,
+    "i": 6,
+    "j": 30,
+    "k": 18,
+    "l": 11,
+    "m": 20,
+    "n": 7,
+    "o": 27,
+    "p": 26,
+    "r": 12,
+    "s": 23,
+    "t": 25,
+    "u": 8,
+    "v": 34,
+    "w": 16,
+    "y": 24,
+    "z": 22,
+    "|": 4,
+    "à": 35,
+    "á": 32,
+    "é": 41,
+    "ì": 38,
+    "í": 31,
+    "ó": 42,
+    "ú": 33,
+    "ń": 36,
+    "ŋ": 15,
+    "ɔ": 40,
+    "ɛ": 9,
+    "́": 37,
+    "ᴐ": 10
+  },
+  "vid": {
+    "'": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 21,
+    "d": 19,
+    "e": 10,
+    "f": 28,
+    "g": 16,
+    "h": 12,
+    "i": 6,
+    "j": 24,
+    "k": 14,
+    "l": 7,
+    "m": 15,
+    "n": 8,
+    "o": 13,
+    "p": 27,
+    "s": 22,
+    "t": 18,
+    "u": 9,
+    "v": 25,
+    "w": 11,
+    "y": 17,
+    "z": 20,
+    "|": 5
+  },
+  "vie": {
+    "!": 132,
+    "\"": 125,
+    "$": 121,
+    "%": 113,
+    "&": 131,
+    "'": 112,
+    "+": 124,
+    ",": 109,
+    "-": 90,
+    ".": 96,
+    "/": 106,
+    "0": 64,
+    "1": 69,
+    "2": 76,
+    "3": 88,
+    "4": 89,
+    "5": 87,
+    "6": 93,
+    "7": 99,
+    "8": 95,
+    "9": 82,
+    ":": 110,
+    ";": 118,
+    "<": 115,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    ">": 116,
+    "[": 128,
+    "]": 129,
+    "a": 11,
+    "b": 21,
+    "c": 8,
+    "d": 27,
+    "e": 31,
+    "f": 80,
+    "g": 10,
+    "h": 6,
+    "i": 9,
+    "j": 92,
+    "k": 24,
+    "l": 19,
+    "m": 15,
+    "n": 5,
+    "o": 14,
+    "p": 26,
+    "q": 46,
+    "r": 17,
+    "s": 25,
+    "t": 7,
+    "u": 12,
+    "v": 18,
+    "w": 81,
+    "x": 56,
+    "y": 22,
+    "z": 97,
+    "|": 4,
+    "¥": 126,
+    "°": 123,
+    "²": 130,
+    "à": 16,
+    "á": 23,
+    "â": 45,
+    "ã": 51,
+    "ç": 127,
+    "è": 102,
+    "é": 84,
+    "ê": 36,
+    "ì": 54,
+    "í": 58,
+    "ò": 71,
+    "ó": 32,
+    "ô": 33,
+    "õ": 108,
+    "ö": 134,
+    "ù": 67,
+    "ú": 63,
+    "ü": 117,
+    "ý": 78,
+    "ă": 57,
+    "đ": 13,
+    "ĩ": 85,
+    "ũ": 77,
+    "ơ": 49,
+    "ư": 20,
+    "ạ": 30,
+    "ả": 34,
+    "ấ": 37,
+    "ầ": 48,
+    "ẩ": 91,
+    "ẫ": 86,
+    "ậ": 50,
+    "ắ": 65,
+    "ằ": 73,
+    "ẳ": 107,
+    "ẵ": 111,
+    "ặ": 61,
+    "ẹ": 105,
+    "ẻ": 98,
+    "ẽ": 74,
+    "ế": 28,
+    "ề": 44,
+    "ể": 38,
+    "ễ": 83,
+    "ệ": 41,
+    "ỉ": 79,
+    "ị": 47,
+    "ọ": 60,
+    "ỏ": 75,
+    "ố": 35,
+    "ồ": 70,
+    "ổ": 66,
+    "ỗ": 94,
+    "ộ": 29,
+    "ớ": 40,
+    "ờ": 39,
+    "ở": 55,
+    "ỡ": 104,
+    "ợ": 43,
+    "ụ": 62,
+    "ủ": 42,
+    "ứ": 59,
+    "ừ": 68,
+    "ử": 72,
+    "ữ": 52,
+    "ự": 53,
+    "ỳ": 100,
+    "ỵ": 119,
+    "ỷ": 101,
+    "ỹ": 103,
+    "‐": 136,
+    "–": 135,
+    "—": 133,
+    "‘": 122,
+    "’": 120,
+    "”": 114
+  },
+  "vif": {
+    "'": 22,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 10,
+    "c": 21,
+    "d": 20,
+    "e": 14,
+    "f": 26,
+    "g": 23,
+    "i": 6,
+    "j": 28,
+    "k": 13,
+    "l": 12,
+    "m": 11,
+    "n": 9,
+    "o": 15,
+    "p": 27,
+    "s": 17,
+    "t": 19,
+    "u": 5,
+    "v": 25,
+    "w": 16,
+    "y": 18,
+    "z": 24,
+    "|": 4,
+    "ê": 31,
+    "ë": 8,
+    "‐": 32,
+    "—": 30
+  },
+  "vmw": {
+    "'": 22,
+    "-": 24,
+    "2": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 27,
+    "c": 25,
+    "d": 29,
+    "e": 8,
+    "f": 26,
+    "g": 31,
+    "h": 10,
+    "i": 6,
+    "j": 28,
+    "k": 15,
+    "l": 16,
+    "m": 14,
+    "n": 9,
+    "o": 7,
+    "p": 20,
+    "r": 17,
+    "s": 19,
+    "t": 13,
+    "u": 11,
+    "v": 21,
+    "w": 12,
+    "x": 23,
+    "y": 18,
+    "|": 5
+  },
+  "vmy": {
+    "'": 12,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 17,
+    "c": 21,
+    "d": 20,
+    "e": 8,
+    "f": 29,
+    "g": 16,
+    "h": 22,
+    "i": 7,
+    "j": 11,
+    "k": 13,
+    "l": 28,
+    "m": 23,
+    "n": 6,
+    "o": 19,
+    "p": 32,
+    "q": 37,
+    "r": 18,
+    "s": 14,
+    "t": 10,
+    "u": 9,
+    "v": 35,
+    "x": 15,
+    "y": 24,
+    "z": 36,
+    "|": 4,
+    "á": 30,
+    "é": 27,
+    "í": 34,
+    "ñ": 26,
+    "ó": 33,
+    "ú": 31,
+    "̱": 25
+  },
+  "vot": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 10,
+    "b": 20,
+    "d": 25,
+    "e": 9,
+    "f": 33,
+    "g": 30,
+    "h": 23,
+    "i": 6,
+    "j": 24,
+    "k": 7,
+    "l": 12,
+    "m": 19,
+    "n": 8,
+    "o": 13,
+    "p": 22,
+    "r": 18,
+    "s": 15,
+    "t": 11,
+    "u": 16,
+    "v": 21,
+    "z": 14,
+    "|": 4,
+    "ä": 17,
+    "õ": 5,
+    "ö": 31,
+    "ü": 26,
+    "č": 27,
+    "š": 32,
+    "ž": 28,
+    "̕": 29
+  },
+  "vun": {
+    "'": 29,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 27,
+    "c": 22,
+    "d": 19,
+    "e": 12,
+    "f": 23,
+    "g": 18,
+    "h": 17,
+    "i": 6,
+    "j": 30,
+    "k": 11,
+    "l": 15,
+    "m": 14,
+    "n": 8,
+    "o": 9,
+    "p": 26,
+    "r": 20,
+    "s": 16,
+    "t": 25,
+    "u": 10,
+    "v": 32,
+    "w": 13,
+    "y": 7,
+    "z": 31,
+    "|": 4,
+    "ṙ": 21,
+    "ṟ": 24
+  },
+  "vut": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 17,
+    "b": 15,
+    "c": 38,
+    "d": 24,
+    "e": 29,
+    "f": 48,
+    "g": 16,
+    "h": 36,
+    "i": 32,
+    "j": 43,
+    "k": 11,
+    "l": 39,
+    "m": 8,
+    "n": 5,
+    "o": 42,
+    "p": 46,
+    "r": 26,
+    "s": 30,
+    "t": 20,
+    "u": 28,
+    "v": 45,
+    "w": 25,
+    "y": 13,
+    "|": 4,
+    "à": 14,
+    "á": 9,
+    "â": 44,
+    "è": 18,
+    "é": 27,
+    "ê": 50,
+    "ì": 33,
+    "í": 10,
+    "î": 49,
+    "ò": 34,
+    "ó": 35,
+    "ô": 51,
+    "ù": 22,
+    "ú": 40,
+    "û": 53,
+    "ŋ": 12,
+    "ȩ": 47,
+    "ɓ": 19,
+    "ɔ": 41,
+    "ɗ": 37,
+    "ə": 23,
+    "ɨ": 6,
+    "̀": 21,
+    "́": 7,
+    "̂": 52,
+    "̧": 31
+  },
+  "wal-script_ethiopic": {
+    "-": 170,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ሀ": 35,
+    "ሁ": 71,
+    "ሂ": 155,
+    "ሃ": 78,
+    "ሄ": 44,
+    "ህ": 95,
+    "ሆ": 136,
+    "ለ": 40,
+    "ሉ": 112,
+    "ሊ": 132,
+    "ላ": 67,
+    "ሌ": 86,
+    "ል": 56,
+    "ሎ": 81,
+    "መ": 22,
+    "ሙ": 96,
+    "ሚ": 109,
+    "ማ": 66,
+    "ሜ": 72,
+    "ም": 51,
+    "ሞ": 87,
+    "ረ": 33,
+    "ሩ": 127,
+    "ሪ": 143,
+    "ራ": 84,
+    "ሬ": 74,
+    "ር": 21,
+    "ሮ": 64,
+    "ሰ": 13,
+    "ሱ": 53,
+    "ሲ": 98,
+    "ሳ": 46,
+    "ሴ": 73,
+    "ስ": 6,
+    "ሶ": 55,
+    "ሸ": 54,
+    "ሹ": 121,
+    "ሺ": 128,
+    "ሻ": 103,
+    "ሼ": 107,
+    "ሽ": 48,
+    "ሾ": 129,
+    "ቀ": 47,
+    "ቁ": 104,
+    "ቂ": 147,
+    "ቃ": 85,
+    "ቄ": 102,
+    "ቅ": 77,
+    "ቆ": 75,
+    "በ": 25,
+    "ቡ": 148,
+    "ቢ": 94,
+    "ባ": 50,
+    "ቤ": 60,
+    "ብ": 76,
+    "ቦ": 69,
+    "ተ": 10,
+    "ቱ": 58,
+    "ቲ": 83,
+    "ታ": 36,
+    "ቴ": 17,
+    "ት": 12,
+    "ቶ": 57,
+    "ቸ": 92,
+    "ቹ": 138,
+    "ቺ": 150,
+    "ቻ": 140,
+    "ቼ": 152,
+    "ች": 106,
+    "ቾ": 145,
+    "ነ": 8,
+    "ኑ": 59,
+    "ኒ": 154,
+    "ና": 41,
+    "ኔ": 20,
+    "ን": 5,
+    "ኖ": 123,
+    "ኙ": 178,
+    "አ": 9,
+    "ኡ": 39,
+    "ኢ": 111,
+    "ኣ": 61,
+    "ኤ": 26,
+    "እ": 11,
+    "ኦ": 32,
+    "ከ": 30,
+    "ኩ": 105,
+    "ኪ": 91,
+    "ካ": 116,
+    "ኬ": 63,
+    "ክ": 62,
+    "ኮ": 37,
+    "ወ": 70,
+    "ዉ": 130,
+    "ዋ": 43,
+    "ዌ": 149,
+    "ው": 27,
+    "ዎ": 31,
+    "ዘ": 80,
+    "ዙ": 141,
+    "ዚ": 160,
+    "ዛ": 126,
+    "ዜ": 124,
+    "ዝ": 119,
+    "ዞ": 135,
+    "የ": 34,
+    "ዩ": 142,
+    "ዪ": 97,
+    "ያ": 14,
+    "ዬ": 42,
+    "ይ": 7,
+    "ዮ": 16,
+    "ደ": 15,
+    "ዱ": 101,
+    "ዲ": 137,
+    "ዳ": 65,
+    "ዴ": 24,
+    "ድ": 18,
+    "ዶ": 23,
+    "ዸ": 162,
+    "ዹ": 180,
+    "ዺ": 163,
+    "ዻ": 179,
+    "ዼ": 168,
+    "ዽ": 114,
+    "ዾ": 176,
+    "ጀ": 167,
+    "ጁ": 166,
+    "ጂ": 175,
+    "ጃ": 174,
+    "ጄ": 151,
+    "ጅ": 164,
+    "ጆ": 172,
+    "ገ": 49,
+    "ጉ": 88,
+    "ጊ": 52,
+    "ጋ": 28,
+    "ጌ": 29,
+    "ግ": 19,
+    "ጎ": 68,
+    "ጠ": 79,
+    "ጡ": 139,
+    "ጢ": 157,
+    "ጣ": 99,
+    "ጤ": 100,
+    "ጥ": 82,
+    "ጦ": 45,
+    "ጨ": 120,
+    "ጩ": 153,
+    "ጪ": 159,
+    "ጫ": 125,
+    "ጬ": 131,
+    "ጭ": 113,
+    "ጮ": 156,
+    "ጰ": 122,
+    "ጱ": 169,
+    "ጲ": 177,
+    "ጳ": 173,
+    "ጴ": 108,
+    "ጵ": 115,
+    "ጶ": 161,
+    "ፈ": 90,
+    "ፉ": 146,
+    "ፊ": 158,
+    "ፋ": 134,
+    "ፌ": 133,
+    "ፍ": 118,
+    "ፎ": 165,
+    "ፐ": 93,
+    "ፑ": 117,
+    "ፒ": 171,
+    "ፓ": 144,
+    "ፔ": 38,
+    "ፕ": 89,
+    "ፖ": 110
+  },
+  "wal-script_latin": {
+    "'": 30,
+    "-": 32,
+    "7": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 27,
+    "d": 12,
+    "e": 7,
+    "f": 29,
+    "g": 16,
+    "h": 14,
+    "i": 6,
+    "j": 31,
+    "k": 17,
+    "l": 22,
+    "m": 19,
+    "n": 10,
+    "o": 9,
+    "p": 21,
+    "q": 24,
+    "r": 20,
+    "s": 8,
+    "t": 11,
+    "u": 15,
+    "w": 23,
+    "x": 25,
+    "y": 13,
+    "z": 28,
+    "|": 5
+  },
+  "wap": {
+    "'": 11,
+    "-": 22,
+    "0": 32,
+    "1": 34,
+    "2": 36,
+    "3": 39,
+    "4": 37,
+    "5": 41,
+    "6": 40,
+    "7": 35,
+    "8": 43,
+    "9": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 28,
+    "d": 13,
+    "e": 24,
+    "f": 31,
+    "g": 30,
+    "h": 20,
+    "i": 6,
+    "j": 27,
+    "k": 10,
+    "l": 29,
+    "m": 16,
+    "n": 7,
+    "o": 9,
+    "p": 12,
+    "q": 42,
+    "r": 15,
+    "s": 19,
+    "t": 14,
+    "u": 8,
+    "v": 33,
+    "w": 18,
+    "x": 38,
+    "y": 26,
+    "z": 17,
+    "|": 5,
+    "ã": 45,
+    "õ": 25,
+    "ĩ": 23
+  },
+  "war": {
+    "'": 37,
+    "-": 24,
+    "0": 32,
+    "1": 34,
+    "2": 35,
+    "3": 39,
+    "4": 33,
+    "5": 38,
+    "6": 44,
+    "7": 42,
+    "8": 40,
+    "9": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 21,
+    "c": 26,
+    "d": 16,
+    "e": 23,
+    "f": 27,
+    "g": 8,
+    "h": 10,
+    "i": 7,
+    "j": 25,
+    "k": 12,
+    "l": 20,
+    "m": 13,
+    "n": 6,
+    "o": 9,
+    "p": 18,
+    "q": 30,
+    "r": 19,
+    "s": 15,
+    "t": 11,
+    "u": 14,
+    "v": 29,
+    "w": 22,
+    "x": 36,
+    "y": 17,
+    "z": 31,
+    "|": 5,
+    "ñ": 43,
+    "—": 28
+  },
+  "waw": {
+    "-": 32,
+    "0": 25,
+    "1": 26,
+    "2": 27,
+    "3": 30,
+    "4": 29,
+    "5": 31,
+    "6": 33,
+    "7": 28,
+    "8": 34,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 22,
+    "d": 36,
+    "e": 7,
+    "h": 15,
+    "i": 19,
+    "k": 8,
+    "m": 11,
+    "n": 9,
+    "o": 6,
+    "p": 18,
+    "r": 10,
+    "s": 20,
+    "t": 12,
+    "u": 23,
+    "w": 14,
+    "x": 21,
+    "y": 16,
+    "|": 4,
+    "î": 13,
+    "̂": 17
+  },
+  "way": {
+    "0": 22,
+    "1": 23,
+    "2": 24,
+    "3": 27,
+    "4": 26,
+    "5": 28,
+    "6": 29,
+    "7": 25,
+    "8": 31,
+    "9": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "e": 8,
+    "h": 17,
+    "i": 14,
+    "j": 20,
+    "k": 11,
+    "l": 15,
+    "m": 7,
+    "n": 12,
+    "o": 10,
+    "p": 13,
+    "s": 21,
+    "t": 9,
+    "u": 18,
+    "w": 19,
+    "|": 4,
+    "ë": 6,
+    "ï": 16
+  },
+  "wba": {
+    "-": 21,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 16,
+    "c": 25,
+    "d": 20,
+    "e": 8,
+    "f": 31,
+    "g": 30,
+    "h": 28,
+    "i": 6,
+    "j": 13,
+    "k": 12,
+    "l": 23,
+    "m": 15,
+    "n": 10,
+    "o": 7,
+    "p": 24,
+    "q": 35,
+    "r": 11,
+    "s": 17,
+    "t": 9,
+    "u": 14,
+    "v": 33,
+    "w": 19,
+    "x": 37,
+    "y": 18,
+    "z": 34,
+    "|": 5,
+    "á": 26,
+    "ã": 36,
+    "é": 27,
+    "í": 32,
+    "ñ": 39,
+    "ó": 29,
+    "õ": 38,
+    "ú": 22
+  },
+  "wlo": {
+    "'": 26,
+    "-": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 17,
+    "c": 23,
+    "d": 21,
+    "e": 11,
+    "f": 29,
+    "g": 16,
+    "h": 18,
+    "i": 6,
+    "j": 27,
+    "k": 12,
+    "l": 20,
+    "m": 8,
+    "n": 7,
+    "o": 10,
+    "p": 15,
+    "r": 19,
+    "s": 14,
+    "t": 13,
+    "u": 9,
+    "w": 22,
+    "y": 28,
+    "z": 30,
+    "|": 5,
+    "ʼ": 25
+  },
+  "wlx": {
+    "'": 12,
+    "-": 27,
+    "1": 34,
+    "4": 33,
+    "5": 32,
+    "9": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 11,
+    "c": 26,
+    "d": 21,
+    "e": 10,
+    "f": 29,
+    "g": 8,
+    "h": 18,
+    "i": 7,
+    "j": 24,
+    "k": 15,
+    "l": 16,
+    "m": 19,
+    "n": 6,
+    "o": 13,
+    "p": 23,
+    "q": 36,
+    "r": 17,
+    "s": 22,
+    "t": 20,
+    "u": 9,
+    "v": 28,
+    "w": 25,
+    "x": 31,
+    "y": 14,
+    "z": 30,
+    "|": 4
+  },
+  "wmw": {
+    "'": 31,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 19,
+    "c": 29,
+    "d": 21,
+    "e": 11,
+    "f": 25,
+    "g": 23,
+    "h": 27,
+    "i": 6,
+    "j": 24,
+    "k": 10,
+    "l": 17,
+    "m": 12,
+    "n": 8,
+    "o": 13,
+    "p": 20,
+    "r": 15,
+    "s": 16,
+    "t": 18,
+    "u": 7,
+    "v": 26,
+    "w": 9,
+    "y": 14,
+    "z": 22,
+    "|": 5,
+    "á": 30,
+    "â": 35,
+    "ã": 37,
+    "é": 34,
+    "ê": 41,
+    "í": 33,
+    "î": 39,
+    "ó": 38,
+    "ô": 42,
+    "ú": 36,
+    "û": 40,
+    "ũ": 43,
+    "‐": 28
+  },
+  "wob": {
+    "'": 5,
+    "-": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 25,
+    "c": 29,
+    "d": 17,
+    "e": 8,
+    "f": 31,
+    "g": 30,
+    "i": 12,
+    "j": 21,
+    "k": 14,
+    "l": 26,
+    "m": 19,
+    "n": 9,
+    "o": 16,
+    "p": 24,
+    "r": 28,
+    "s": 18,
+    "t": 23,
+    "u": 22,
+    "v": 33,
+    "w": 20,
+    "y": 27,
+    "|": 4,
+    "ɔ": 13,
+    "ɛ": 10,
+    "ɩ": 15,
+    "ʋ": 11,
+    "‐": 7
+  },
+  "wol": {
+    "\"": 74,
+    "$": 67,
+    "%": 61,
+    "&": 77,
+    "'": 48,
+    "+": 80,
+    ",": 49,
+    "-": 32,
+    ".": 47,
+    "/": 54,
+    "0": 34,
+    "1": 36,
+    "2": 38,
+    "3": 42,
+    "4": 44,
+    "5": 41,
+    "6": 43,
+    "7": 51,
+    "8": 46,
+    "9": 45,
+    ":": 52,
+    ";": 63,
+    "<": 72,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    ">": 73,
+    "?": 76,
+    "[": 84,
+    "]": 85,
+    "a": 5,
+    "b": 15,
+    "c": 21,
+    "d": 18,
+    "e": 8,
+    "f": 28,
+    "g": 19,
+    "h": 31,
+    "i": 6,
+    "j": 23,
+    "k": 14,
+    "l": 11,
+    "m": 17,
+    "n": 7,
+    "o": 10,
+    "p": 25,
+    "q": 37,
+    "r": 16,
+    "s": 20,
+    "t": 13,
+    "u": 9,
+    "v": 33,
+    "w": 22,
+    "x": 24,
+    "y": 12,
+    "z": 39,
+    "|": 4,
+    "«": 59,
+    "°": 81,
+    "²": 75,
+    "»": 60,
+    "à": 30,
+    "á": 70,
+    "â": 64,
+    "ã": 69,
+    "ç": 57,
+    "è": 50,
+    "é": 29,
+    "ê": 55,
+    "ë": 27,
+    "í": 71,
+    "î": 82,
+    "ï": 62,
+    "ñ": 26,
+    "ó": 35,
+    "ô": 66,
+    "õ": 68,
+    "ù": 86,
+    "ú": 78,
+    "ü": 58,
+    "ŋ": 40,
+    "–": 79,
+    "—": 83,
+    "‘": 65,
+    "’": 53,
+    "”": 56
+  },
+  "wsg": {
+    "'": 41,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "k": 44,
+    "|": 4,
+    "ం": 16,
+    "అ": 27,
+    "ఆ": 34,
+    "ఇ": 30,
+    "ఈ": 39,
+    "ఉ": 38,
+    "ఊ": 42,
+    "ఎ": 40,
+    "ఒ": 43,
+    "క": 11,
+    "గ": 21,
+    "చ": 36,
+    "జ": 29,
+    "ట": 25,
+    "డ": 26,
+    "త": 9,
+    "ద": 20,
+    "న": 8,
+    "ప": 24,
+    "బ": 23,
+    "మ": 14,
+    "య": 22,
+    "ర": 6,
+    "ల": 17,
+    "వ": 12,
+    "స": 13,
+    "హ": 31,
+    "ా": 19,
+    "ి": 7,
+    "ీ": 28,
+    "ు": 15,
+    "ూ": 35,
+    "ె": 10,
+    "ే": 33,
+    "ొ": 18,
+    "ో": 32,
+    "్": 5
+  },
+  "wwa": {
+    "'": 34,
+    "-": 27,
+    "2": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "`": 35,
+    "a": 5,
+    "b": 12,
+    "c": 30,
+    "d": 9,
+    "e": 14,
+    "f": 29,
+    "i": 6,
+    "k": 18,
+    "m": 11,
+    "n": 7,
+    "o": 10,
+    "p": 23,
+    "r": 13,
+    "s": 15,
+    "t": 8,
+    "u": 16,
+    "w": 22,
+    "y": 17,
+    "|": 4,
+    "à": 21,
+    "ã": 32,
+    "ì": 28,
+    "ò": 24,
+    "ĩ": 36,
+    "ŋ": 25,
+    "ũ": 37,
+    "ǹ": 26,
+    "ɔ": 20,
+    "ɛ": 19,
+    "̀": 33,
+    "̃": 31
+  },
+  "xal": {
+    "-": 38,
+    "0": 47,
+    "1": 49,
+    "2": 50,
+    "3": 46,
+    "4": 48,
+    "6": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "а": 6,
+    "б": 19,
+    "в": 22,
+    "г": 13,
+    "д": 7,
+    "е": 17,
+    "ж": 45,
+    "з": 27,
+    "и": 11,
+    "й": 32,
+    "к": 18,
+    "л": 10,
+    "м": 23,
+    "н": 5,
+    "о": 21,
+    "п": 39,
+    "р": 9,
+    "с": 14,
+    "т": 12,
+    "у": 15,
+    "ф": 43,
+    "х": 16,
+    "ц": 28,
+    "ч": 26,
+    "ш": 33,
+    "ъ": 44,
+    "ы": 41,
+    "ь": 30,
+    "э": 29,
+    "ю": 40,
+    "я": 37,
+    "ё": 42,
+    "ѳ": 31,
+    "җ": 24,
+    "ң": 34,
+    "ү": 20,
+    "һ": 25,
+    "ә": 8,
+    "ө": 35,
+    "–": 36
+  },
+  "xdy": {
+    "'": 14,
+    "-": 26,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 25,
+    "d": 16,
+    "e": 24,
+    "f": 28,
+    "g": 19,
+    "h": 22,
+    "i": 8,
+    "j": 23,
+    "k": 10,
+    "l": 18,
+    "m": 12,
+    "n": 7,
+    "o": 6,
+    "p": 17,
+    "r": 21,
+    "s": 13,
+    "t": 11,
+    "u": 9,
+    "w": 27,
+    "y": 20,
+    "z": 29,
+    "|": 4
+  },
+  "xed": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 28,
+    "d": 15,
+    "e": 29,
+    "f": 23,
+    "g": 11,
+    "h": 12,
+    "i": 10,
+    "k": 8,
+    "l": 13,
+    "m": 14,
+    "n": 7,
+    "p": 30,
+    "r": 24,
+    "s": 16,
+    "t": 6,
+    "u": 9,
+    "v": 27,
+    "w": 17,
+    "y": 19,
+    "z": 20,
+    "|": 5,
+    "á": 25,
+    "í": 32,
+    "ŋ": 18,
+    "ɓ": 31,
+    "ɗ": 22,
+    "ə": 21,
+    "ʼ": 26
+  },
+  "xer": {
+    "'": 50,
+    "-": 54,
+    "0": 37,
+    "1": 35,
+    "2": 41,
+    "3": 53,
+    "5": 42,
+    "6": 52,
+    "7": 45,
+    "8": 48,
+    "9": 47,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 36,
+    "d": 15,
+    "e": 16,
+    "f": 33,
+    "g": 34,
+    "h": 25,
+    "i": 13,
+    "j": 29,
+    "k": 6,
+    "l": 31,
+    "m": 7,
+    "n": 11,
+    "o": 26,
+    "p": 21,
+    "q": 43,
+    "r": 8,
+    "s": 14,
+    "t": 10,
+    "u": 18,
+    "v": 38,
+    "w": 12,
+    "x": 51,
+    "z": 19,
+    "|": 4,
+    "á": 40,
+    "â": 24,
+    "ã": 9,
+    "é": 32,
+    "ê": 27,
+    "í": 49,
+    "ó": 44,
+    "ô": 28,
+    "õ": 17,
+    "ø": 46,
+    "ú": 55,
+    "û": 39,
+    "ĩ": 22,
+    "ũ": 56,
+    "ẽ": 20,
+    "—": 30
+  },
+  "xho": {
+    "!": 75,
+    "\"": 52,
+    "$": 47,
+    "%": 51,
+    "&": 58,
+    "'": 44,
+    "+": 60,
+    ",": 43,
+    "-": 26,
+    ".": 38,
+    "/": 46,
+    "0": 32,
+    "1": 33,
+    "2": 34,
+    "3": 39,
+    "4": 37,
+    "5": 36,
+    "6": 40,
+    "7": 42,
+    "8": 41,
+    "9": 35,
+    ":": 45,
+    ";": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 66,
+    "]": 67,
+    "a": 4,
+    "b": 15,
+    "c": 25,
+    "d": 22,
+    "e": 6,
+    "f": 28,
+    "g": 21,
+    "h": 13,
+    "i": 7,
+    "j": 31,
+    "k": 10,
+    "l": 12,
+    "m": 18,
+    "n": 8,
+    "o": 11,
+    "p": 23,
+    "q": 27,
+    "r": 24,
+    "s": 14,
+    "t": 20,
+    "u": 9,
+    "v": 30,
+    "w": 17,
+    "x": 29,
+    "y": 16,
+    "z": 19,
+    "|": 5,
+    "£": 68,
+    "¥": 56,
+    "°": 59,
+    "²": 69,
+    "á": 50,
+    "ã": 63,
+    "ç": 64,
+    "é": 61,
+    "í": 54,
+    "ó": 72,
+    "õ": 55,
+    "ö": 74,
+    "ú": 62,
+    "ü": 53,
+    "̇": 71,
+    "–": 65,
+    "—": 57,
+    "‘": 70,
+    "’": 73,
+    "”": 49
+  },
+  "xmm": {
+    "'": 29,
+    "-": 24,
+    "0": 30,
+    "1": 34,
+    "2": 33,
+    "3": 35,
+    "4": 31,
+    "6": 32,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 20,
+    "c": 25,
+    "d": 14,
+    "e": 13,
+    "f": 26,
+    "g": 9,
+    "h": 22,
+    "i": 7,
+    "j": 23,
+    "k": 17,
+    "l": 18,
+    "m": 19,
+    "n": 6,
+    "o": 11,
+    "p": 16,
+    "r": 12,
+    "s": 15,
+    "t": 10,
+    "u": 8,
+    "w": 27,
+    "y": 21,
+    "z": 28,
+    "|": 4
+  },
+  "xnj": {
+    "'": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 22,
+    "c": 24,
+    "d": 20,
+    "e": 12,
+    "f": 27,
+    "g": 14,
+    "h": 18,
+    "i": 6,
+    "j": 25,
+    "k": 11,
+    "l": 9,
+    "m": 10,
+    "n": 8,
+    "o": 15,
+    "p": 21,
+    "s": 23,
+    "t": 19,
+    "u": 7,
+    "v": 13,
+    "w": 16,
+    "y": 17,
+    "z": 26,
+    "|": 5
+  },
+  "xnr": {
+    "'": 62,
+    "-": 51,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 58,
+    "ं": 8,
+    "अ": 31,
+    "आ": 32,
+    "इ": 37,
+    "ई": 33,
+    "उ": 55,
+    "ऊ": 54,
+    "ए": 42,
+    "ऐ": 35,
+    "ओ": 56,
+    "औ": 59,
+    "क": 11,
+    "ख": 38,
+    "ग": 27,
+    "घ": 52,
+    "च": 30,
+    "छ": 46,
+    "ज": 18,
+    "झ": 50,
+    "ञ": 60,
+    "ट": 49,
+    "ठ": 48,
+    "ड": 34,
+    "ढ": 57,
+    "ण": 28,
+    "त": 9,
+    "थ": 40,
+    "द": 15,
+    "ध": 45,
+    "न": 16,
+    "प": 22,
+    "फ": 47,
+    "ब": 29,
+    "भ": 39,
+    "म": 19,
+    "य": 21,
+    "र": 10,
+    "ल": 20,
+    "व": 41,
+    "श": 43,
+    "ष": 53,
+    "स": 14,
+    "ह": 12,
+    "़": 26,
+    "ा": 5,
+    "ि": 13,
+    "ी": 17,
+    "ु": 24,
+    "ू": 36,
+    "ृ": 61,
+    "े": 6,
+    "ै": 25,
+    "ो": 23,
+    "ौ": 44,
+    "्": 7,
+    "—": 63
+  },
+  "xog": {
+    "'": 23,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 10,
+    "d": 22,
+    "e": 7,
+    "f": 25,
+    "g": 14,
+    "h": 21,
+    "i": 6,
+    "j": 28,
+    "k": 12,
+    "l": 16,
+    "m": 13,
+    "n": 11,
+    "o": 9,
+    "p": 27,
+    "r": 19,
+    "s": 20,
+    "t": 18,
+    "u": 8,
+    "v": 26,
+    "w": 17,
+    "y": 15,
+    "z": 24,
+    "|": 5
+  },
+  "xon": {
+    "'": 30,
+    "-": 29,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 27,
+    "d": 24,
+    "e": 11,
+    "f": 28,
+    "g": 23,
+    "h": 22,
+    "i": 6,
+    "j": 26,
+    "k": 10,
+    "l": 13,
+    "m": 12,
+    "n": 7,
+    "o": 17,
+    "p": 16,
+    "r": 18,
+    "s": 21,
+    "t": 14,
+    "u": 8,
+    "v": 31,
+    "w": 20,
+    "y": 19,
+    "|": 4,
+    "ŋ": 25,
+    "ɔ": 15
+  },
+  "xrb": {
+    "'": 35,
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 31,
+    "d": 28,
+    "e": 6,
+    "f": 27,
+    "g": 22,
+    "h": 17,
+    "i": 10,
+    "j": 36,
+    "k": 13,
+    "l": 18,
+    "m": 16,
+    "n": 8,
+    "o": 11,
+    "p": 15,
+    "r": 21,
+    "s": 19,
+    "t": 20,
+    "u": 14,
+    "w": 12,
+    "y": 7,
+    "|": 4,
+    "ã": 9,
+    "ĩ": 23,
+    "ŋ": 29,
+    "ũ": 32,
+    "ɔ": 33,
+    "ɛ": 25,
+    "̃": 34,
+    "ẽ": 26
+  },
+  "xsb": {
+    "'": 28,
+    "-": 19,
+    "0": 32,
+    "1": 36,
+    "2": 37,
+    "4": 43,
+    "6": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 26,
+    "d": 22,
+    "e": 23,
+    "f": 27,
+    "g": 15,
+    "h": 16,
+    "i": 8,
+    "j": 25,
+    "k": 12,
+    "l": 13,
+    "m": 10,
+    "n": 6,
+    "o": 7,
+    "p": 14,
+    "q": 31,
+    "r": 21,
+    "s": 17,
+    "t": 9,
+    "u": 24,
+    "v": 29,
+    "w": 20,
+    "x": 35,
+    "y": 11,
+    "z": 30,
+    "|": 5,
+    "à": 40,
+    "á": 38,
+    "â": 34,
+    "î": 44,
+    "ñ": 41,
+    "ó": 42,
+    "ô": 33
+  },
+  "xsm": {
+    "'": 36,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 31,
+    "d": 14,
+    "e": 6,
+    "f": 35,
+    "g": 20,
+    "h": 30,
+    "i": 11,
+    "j": 28,
+    "k": 15,
+    "l": 19,
+    "m": 9,
+    "n": 8,
+    "o": 7,
+    "p": 26,
+    "r": 18,
+    "s": 21,
+    "t": 10,
+    "u": 23,
+    "v": 34,
+    "w": 12,
+    "y": 16,
+    "z": 29,
+    "|": 4,
+    "á": 27,
+    "é": 32,
+    "ó": 33,
+    "ŋ": 22,
+    "ǝ": 37,
+    "ɔ": 17,
+    "ɛ": 24
+  },
+  "xsr": {
+    "'": 49,
+    "-": 58,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "ँ": 48,
+    "ं": 50,
+    "अ": 41,
+    "आ": 44,
+    "इ": 42,
+    "ई": 54,
+    "उ": 46,
+    "ऊ": 47,
+    "ए": 45,
+    "ओ": 39,
+    "क": 8,
+    "ख": 25,
+    "ग": 26,
+    "घ": 55,
+    "ङ": 13,
+    "च": 28,
+    "छ": 29,
+    "ज": 27,
+    "झ": 59,
+    "ट": 33,
+    "ठ": 38,
+    "ड": 36,
+    "ण": 52,
+    "त": 11,
+    "थ": 31,
+    "द": 24,
+    "ध": 57,
+    "न": 16,
+    "प": 20,
+    "फ": 37,
+    "ब": 30,
+    "भ": 53,
+    "म": 19,
+    "य": 14,
+    "र": 23,
+    "ल": 17,
+    "व": 22,
+    "श": 32,
+    "ष": 40,
+    "स": 21,
+    "ह": 34,
+    "ा": 5,
+    "ि": 9,
+    "ी": 10,
+    "ु": 12,
+    "ू": 35,
+    "े": 15,
+    "ै": 43,
+    "ो": 18,
+    "ौ": 51,
+    "्": 6,
+    "६": 56,
+    "‍": 7
+  },
+  "xsu": {
+    "'": 22,
+    "0": 32,
+    "1": 38,
+    "2": 42,
+    "3": 43,
+    "4": 41,
+    "5": 45,
+    "6": 50,
+    "7": 40,
+    "8": 51,
+    "9": 53,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 28,
+    "d": 31,
+    "e": 18,
+    "f": 46,
+    "g": 48,
+    "h": 19,
+    "i": 7,
+    "j": 23,
+    "k": 10,
+    "l": 16,
+    "m": 15,
+    "n": 12,
+    "o": 11,
+    "p": 9,
+    "r": 26,
+    "s": 13,
+    "t": 8,
+    "u": 14,
+    "v": 36,
+    "w": 21,
+    "|": 4,
+    "á": 35,
+    "â": 52,
+    "ã": 25,
+    "ä": 20,
+    "é": 29,
+    "ê": 44,
+    "í": 39,
+    "ó": 33,
+    "ô": 37,
+    "õ": 24,
+    "ö": 6,
+    "ú": 49,
+    "ĩ": 17,
+    "ũ": 47,
+    "̃": 27,
+    "ẽ": 30
+  },
+  "xta": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 31,
+    "c": 21,
+    "d": 11,
+    "e": 17,
+    "f": 36,
+    "g": 35,
+    "h": 23,
+    "i": 7,
+    "j": 32,
+    "k": 9,
+    "l": 25,
+    "m": 30,
+    "n": 6,
+    "o": 13,
+    "p": 29,
+    "q": 39,
+    "r": 15,
+    "s": 16,
+    "t": 12,
+    "u": 8,
+    "v": 22,
+    "x": 14,
+    "y": 20,
+    "z": 41,
+    "|": 4,
+    "á": 19,
+    "ä": 38,
+    "é": 34,
+    "ë": 43,
+    "í": 26,
+    "ï": 40,
+    "ñ": 24,
+    "ó": 28,
+    "ö": 33,
+    "ú": 27,
+    "ü": 37,
+    "ǎ": 44,
+    "ǔ": 42,
+    "̱": 18,
+    "ꞌ": 10
+  },
+  "xtd": {
+    "'": 14,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 34,
+    "c": 23,
+    "d": 13,
+    "e": 12,
+    "f": 37,
+    "g": 31,
+    "h": 26,
+    "i": 8,
+    "j": 19,
+    "k": 10,
+    "l": 27,
+    "m": 24,
+    "n": 6,
+    "o": 16,
+    "p": 29,
+    "q": 40,
+    "r": 22,
+    "s": 17,
+    "t": 15,
+    "u": 7,
+    "v": 25,
+    "x": 11,
+    "y": 20,
+    "z": 38,
+    "|": 4,
+    "á": 28,
+    "é": 35,
+    "í": 33,
+    "ñ": 18,
+    "ó": 36,
+    "ú": 30,
+    "ɨ": 21,
+    "ʼ": 39,
+    "́": 41,
+    "‐": 9,
+    "—": 32
+  },
+  "xte": {
+    "-": 25,
+    "0": 28,
+    "1": 29,
+    "2": 30,
+    "3": 34,
+    "4": 32,
+    "5": 33,
+    "6": 37,
+    "7": 35,
+    "8": 38,
+    "9": 36,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 22,
+    "d": 12,
+    "e": 6,
+    "f": 27,
+    "g": 19,
+    "h": 24,
+    "i": 10,
+    "j": 26,
+    "k": 13,
+    "l": 18,
+    "m": 11,
+    "n": 7,
+    "o": 16,
+    "p": 9,
+    "r": 14,
+    "s": 17,
+    "t": 15,
+    "u": 8,
+    "w": 23,
+    "y": 20,
+    "z": 31,
+    "|": 4
+  },
+  "xtm": {
+    "'": 11,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 33,
+    "c": 21,
+    "d": 17,
+    "e": 10,
+    "f": 39,
+    "g": 36,
+    "h": 24,
+    "i": 6,
+    "j": 13,
+    "k": 9,
+    "l": 28,
+    "m": 15,
+    "n": 7,
+    "o": 16,
+    "p": 31,
+    "q": 41,
+    "r": 27,
+    "s": 14,
+    "t": 12,
+    "u": 8,
+    "v": 20,
+    "w": 32,
+    "x": 25,
+    "y": 18,
+    "z": 40,
+    "|": 4,
+    "á": 26,
+    "â": 35,
+    "é": 34,
+    "ê": 42,
+    "í": 30,
+    "î": 43,
+    "ñ": 19,
+    "ó": 37,
+    "ú": 22,
+    "û": 38,
+    "‍": 29,
+    "‐": 23
+  },
+  "xtn": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 30,
+    "c": 20,
+    "d": 16,
+    "e": 10,
+    "f": 33,
+    "g": 25,
+    "h": 21,
+    "i": 6,
+    "j": 14,
+    "k": 9,
+    "l": 26,
+    "m": 22,
+    "n": 7,
+    "o": 13,
+    "p": 27,
+    "q": 37,
+    "r": 23,
+    "s": 15,
+    "t": 11,
+    "u": 8,
+    "v": 19,
+    "x": 24,
+    "y": 17,
+    "z": 35,
+    "|": 4,
+    "á": 36,
+    "é": 32,
+    "í": 31,
+    "ñ": 18,
+    "ó": 34,
+    "ú": 28,
+    "͏": 29,
+    "ꞌ": 12
+  },
+  "xua": {
+    "-": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 4,
+    "அ": 23,
+    "ஆ": 29,
+    "இ": 27,
+    "ஈ": 41,
+    "உ": 40,
+    "ஊ": 45,
+    "எ": 26,
+    "ஏ": 25,
+    "ஐ": 47,
+    "ஒ": 32,
+    "ஓ": 37,
+    "க": 8,
+    "ங": 28,
+    "ச": 19,
+    "ஜ": 33,
+    "ஞ": 43,
+    "ட": 11,
+    "ண": 31,
+    "த": 7,
+    "ந": 17,
+    "ன": 14,
+    "ப": 13,
+    "ம": 21,
+    "ய": 20,
+    "ர": 12,
+    "ற": 44,
+    "ல": 18,
+    "ள": 35,
+    "ழ": 46,
+    "வ": 15,
+    "ஸ": 38,
+    "ஹ": 39,
+    "ா": 9,
+    "ி": 10,
+    "ீ": 36,
+    "ு": 5,
+    "ூ": 34,
+    "ெ": 16,
+    "ே": 24,
+    "ை": 42,
+    "ொ": 30,
+    "ோ": 22,
+    "்": 6
+  },
+  "xuo": {
+    "'": 53,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "d": 34,
+    "e": 12,
+    "f": 30,
+    "g": 31,
+    "h": 27,
+    "i": 6,
+    "k": 7,
+    "l": 14,
+    "m": 17,
+    "n": 10,
+    "o": 15,
+    "p": 35,
+    "r": 8,
+    "s": 22,
+    "t": 26,
+    "u": 11,
+    "v": 38,
+    "w": 21,
+    "y": 13,
+    "z": 19,
+    "|": 4,
+    "à": 51,
+    "á": 16,
+    "é": 44,
+    "ì": 29,
+    "í": 37,
+    "ó": 47,
+    "ù": 39,
+    "ú": 41,
+    "ě": 52,
+    "ŋ": 24,
+    "ǎ": 42,
+    "ǐ": 49,
+    "ǒ": 50,
+    "ǔ": 48,
+    "ɓ": 9,
+    "ɔ": 25,
+    "ɗ": 20,
+    "ɛ": 18,
+    "ʼ": 43,
+    "̀": 45,
+    "́": 40,
+    "̌": 46,
+    "̰": 28,
+    "ḭ": 32,
+    "ṵ": 36
+  },
+  "yaa": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 30,
+    "c": 24,
+    "d": 29,
+    "e": 12,
+    "f": 10,
+    "g": 34,
+    "h": 20,
+    "i": 6,
+    "j": 25,
+    "k": 8,
+    "l": 28,
+    "m": 13,
+    "n": 9,
+    "o": 7,
+    "p": 19,
+    "q": 38,
+    "r": 16,
+    "s": 11,
+    "t": 15,
+    "u": 27,
+    "v": 35,
+    "x": 18,
+    "y": 21,
+    "z": 36,
+    "|": 5,
+    "á": 33,
+    "ã": 14,
+    "é": 31,
+    "í": 32,
+    "ñ": 39,
+    "ó": 37,
+    "õ": 17,
+    "ú": 26,
+    "ĩ": 22,
+    "ẽ": 23
+  },
+  "yad": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 23,
+    "c": 19,
+    "d": 14,
+    "e": 17,
+    "g": 32,
+    "h": 20,
+    "i": 6,
+    "j": 11,
+    "l": 31,
+    "m": 18,
+    "n": 13,
+    "o": 21,
+    "p": 24,
+    "q": 25,
+    "r": 10,
+    "s": 15,
+    "t": 12,
+    "u": 9,
+    "v": 16,
+    "y": 7,
+    "|": 5,
+    "á": 27,
+    "é": 29,
+    "í": 28,
+    "ñ": 22,
+    "ó": 30,
+    "ú": 26,
+    "̱": 8
+  },
+  "yal": {
+    "-": 29,
+    "1": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 31,
+    "d": 22,
+    "e": 8,
+    "f": 18,
+    "g": 24,
+    "h": 27,
+    "i": 7,
+    "k": 23,
+    "l": 14,
+    "m": 11,
+    "n": 6,
+    "o": 20,
+    "p": 28,
+    "r": 16,
+    "s": 17,
+    "t": 15,
+    "u": 13,
+    "w": 26,
+    "x": 9,
+    "y": 10,
+    "|": 4,
+    "ɔ": 21,
+    "ɛ": 12,
+    "ɲ": 25
+  },
+  "yam": {
+    "'": 18,
+    "-": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 20,
+    "c": 33,
+    "d": 24,
+    "e": 16,
+    "f": 27,
+    "g": 22,
+    "h": 37,
+    "i": 21,
+    "j": 38,
+    "k": 15,
+    "l": 31,
+    "m": 12,
+    "n": 6,
+    "o": 17,
+    "p": 30,
+    "r": 36,
+    "s": 7,
+    "t": 25,
+    "u": 28,
+    "v": 29,
+    "w": 11,
+    "y": 13,
+    "z": 35,
+    "|": 4,
+    "à": 14,
+    "è": 23,
+    "ì": 32,
+    "ò": 26,
+    "ù": 39,
+    "ŋ": 10,
+    "ə": 5,
+    "ɛ": 34,
+    "ʉ": 19,
+    "ʼ": 40,
+    "ˋ": 42,
+    "̀": 8,
+    "̍": 43
+  },
+  "yao": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 24,
+    "c": 20,
+    "d": 25,
+    "e": 9,
+    "f": 26,
+    "g": 15,
+    "h": 18,
+    "i": 6,
+    "j": 23,
+    "k": 11,
+    "l": 10,
+    "m": 13,
+    "n": 7,
+    "o": 12,
+    "p": 21,
+    "r": 30,
+    "s": 16,
+    "t": 22,
+    "u": 8,
+    "v": 28,
+    "w": 14,
+    "y": 19,
+    "z": 29,
+    "|": 5,
+    "ŵ": 17
+  },
+  "yas": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 10,
+    "c": 30,
+    "d": 16,
+    "e": 15,
+    "f": 31,
+    "g": 12,
+    "h": 29,
+    "i": 19,
+    "k": 27,
+    "l": 18,
+    "m": 13,
+    "n": 7,
+    "o": 14,
+    "p": 32,
+    "s": 21,
+    "t": 28,
+    "u": 24,
+    "y": 17,
+    "|": 4,
+    "á": 11,
+    "é": 25,
+    "í": 22,
+    "ó": 23,
+    "ú": 26,
+    "ŋ": 20,
+    "ɔ": 6,
+    "ɛ": 9,
+    "́": 8,
+    "–": 33
+  },
+  "yat": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 27,
+    "c": 33,
+    "d": 19,
+    "e": 35,
+    "f": 31,
+    "g": 20,
+    "i": 18,
+    "j": 32,
+    "k": 14,
+    "l": 13,
+    "m": 12,
+    "n": 7,
+    "o": 16,
+    "p": 15,
+    "r": 36,
+    "s": 21,
+    "t": 22,
+    "u": 24,
+    "w": 30,
+    "y": 17,
+    "|": 4,
+    "á": 10,
+    "è": 37,
+    "é": 34,
+    "í": 28,
+    "ó": 23,
+    "ú": 29,
+    "ŋ": 26,
+    "ɔ": 9,
+    "ə": 11,
+    "ɛ": 5,
+    "ʼ": 25,
+    "́": 6
+  },
+  "yaz": {
+    "'": 28,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 35,
+    "d": 19,
+    "e": 16,
+    "f": 26,
+    "g": 22,
+    "h": 32,
+    "i": 10,
+    "j": 24,
+    "k": 11,
+    "l": 17,
+    "m": 15,
+    "n": 7,
+    "o": 8,
+    "p": 25,
+    "r": 27,
+    "s": 23,
+    "t": 21,
+    "u": 20,
+    "v": 31,
+    "w": 18,
+    "y": 12,
+    "z": 30,
+    "|": 4,
+    "à": 33,
+    "ì": 34,
+    "ò": 29,
+    "̀": 36,
+    "ạ": 6,
+    "ẹ": 9,
+    "ọ": 14
+  },
+  "yba": {
+    "'": 37,
+    "-": 35,
+    "6": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 24,
+    "d": 29,
+    "e": 15,
+    "f": 36,
+    "g": 18,
+    "h": 12,
+    "i": 9,
+    "j": 31,
+    "k": 19,
+    "l": 10,
+    "m": 21,
+    "n": 13,
+    "o": 23,
+    "p": 28,
+    "r": 30,
+    "s": 33,
+    "t": 32,
+    "u": 25,
+    "w": 22,
+    "y": 14,
+    "|": 4,
+    "ā": 16,
+    "ē": 26,
+    "ī": 11,
+    "ō": 34,
+    "ū": 27,
+    "ɔ": 7,
+    "ɛ": 8,
+    "̄": 17,
+    "̍": 5
+  },
+  "ybb": {
+    "-": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 13,
+    "b": 33,
+    "c": 43,
+    "d": 30,
+    "e": 5,
+    "f": 36,
+    "g": 20,
+    "h": 23,
+    "i": 21,
+    "j": 38,
+    "k": 22,
+    "l": 19,
+    "m": 9,
+    "n": 11,
+    "o": 39,
+    "p": 15,
+    "s": 18,
+    "t": 14,
+    "u": 31,
+    "v": 47,
+    "w": 25,
+    "y": 28,
+    "z": 26,
+    "|": 4,
+    "á": 10,
+    "é": 16,
+    "í": 29,
+    "ó": 42,
+    "ú": 34,
+    "ā": 37,
+    "ē": 41,
+    "ī": 32,
+    "ń": 40,
+    "ŋ": 7,
+    "ō": 44,
+    "ū": 35,
+    "ɔ": 6,
+    "ɛ": 8,
+    "ʉ": 24,
+    "ʼ": 17,
+    "́": 12,
+    "̄": 27,
+    "ḿ": 45,
+    "‐": 46
+  },
+  "ycl": {
+    "'": 33,
+    "-": 24,
+    "0": 32,
+    "1": 34,
+    "2": 35,
+    "3": 37,
+    "4": 36,
+    "5": 40,
+    "6": 38,
+    "9": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 21,
+    "d": 10,
+    "e": 6,
+    "f": 30,
+    "g": 16,
+    "h": 22,
+    "i": 7,
+    "j": 23,
+    "k": 28,
+    "l": 8,
+    "m": 17,
+    "n": 13,
+    "o": 12,
+    "p": 27,
+    "q": 31,
+    "r": 9,
+    "s": 14,
+    "t": 26,
+    "u": 19,
+    "v": 25,
+    "w": 29,
+    "x": 11,
+    "y": 20,
+    "z": 18,
+    "|": 4
+  },
+  "ycn": {
+    "'": 11,
+    "-": 40,
+    "0": 39,
+    "1": 43,
+    "4": 41,
+    "6": 42,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 32,
+    "c": 23,
+    "d": 30,
+    "e": 8,
+    "f": 33,
+    "g": 34,
+    "h": 21,
+    "i": 7,
+    "j": 9,
+    "k": 6,
+    "l": 16,
+    "m": 18,
+    "n": 10,
+    "o": 17,
+    "p": 15,
+    "q": 37,
+    "r": 13,
+    "s": 29,
+    "t": 20,
+    "u": 14,
+    "v": 36,
+    "w": 19,
+    "x": 38,
+    "y": 24,
+    "z": 35,
+    "|": 5,
+    "á": 12,
+    "é": 22,
+    "í": 28,
+    "ñ": 26,
+    "ó": 27,
+    "ú": 25,
+    "—": 31
+  },
+  "yea": {
+    "'": 46,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "|": 5,
+    "അ": 24,
+    "ആ": 39,
+    "ഇ": 37,
+    "ഈ": 35,
+    "ഉ": 42,
+    "ഊ": 44,
+    "എ": 31,
+    "ഏ": 38,
+    "ഒ": 36,
+    "ഓ": 43,
+    "ക": 8,
+    "ഗ": 33,
+    "ങ": 26,
+    "ച": 11,
+    "ജ": 41,
+    "ഞ": 34,
+    "ട": 15,
+    "ഡ": 45,
+    "ണ": 22,
+    "ത": 13,
+    "ദ": 28,
+    "ന": 7,
+    "പ": 17,
+    "ബ": 25,
+    "മ": 19,
+    "യ": 16,
+    "ര": 18,
+    "റ": 12,
+    "ല": 20,
+    "ള": 29,
+    "വ": 27,
+    "ാ": 14,
+    "ി": 9,
+    "ീ": 32,
+    "ു": 6,
+    "ൂ": 40,
+    "െ": 10,
+    "േ": 21,
+    "ൊ": 23,
+    "ോ": 30,
+    "്": 4,
+    "ൺ": 47,
+    "‍": 48,
+    "—": 49
+  },
+  "yka": {
+    "'": 32,
+    "-": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 34,
+    "d": 17,
+    "e": 8,
+    "g": 14,
+    "h": 20,
+    "i": 7,
+    "j": 25,
+    "k": 12,
+    "l": 18,
+    "m": 13,
+    "n": 6,
+    "o": 23,
+    "p": 19,
+    "r": 24,
+    "s": 11,
+    "t": 15,
+    "u": 9,
+    "w": 22,
+    "y": 21,
+    "|": 4,
+    "ã": 33,
+    "ā": 26,
+    "ē": 28,
+    "ī": 29,
+    "ō": 30,
+    "ū": 31,
+    "ꞌ": 10
+  },
+  "yli": {
+    "'": 26,
+    "-": 25,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 29,
+    "d": 24,
+    "e": 7,
+    "f": 22,
+    "g": 12,
+    "h": 11,
+    "i": 8,
+    "j": 28,
+    "k": 17,
+    "l": 13,
+    "m": 14,
+    "n": 6,
+    "o": 10,
+    "p": 20,
+    "r": 16,
+    "s": 18,
+    "t": 15,
+    "u": 9,
+    "w": 19,
+    "y": 23,
+    "z": 27,
+    "|": 4
+  },
+  "yor": {
+    "!": 88,
+    "$": 80,
+    "%": 74,
+    "'": 62,
+    "+": 86,
+    ",": 56,
+    "-": 44,
+    ".": 46,
+    "/": 67,
+    "0": 43,
+    "1": 45,
+    "2": 48,
+    "3": 57,
+    "4": 53,
+    "5": 54,
+    "6": 55,
+    "7": 59,
+    "8": 52,
+    "9": 50,
+    ":": 63,
+    ";": 66,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "=": 91,
+    "?": 72,
+    "[": 83,
+    "\\": 78,
+    "]": 84,
+    "a": 7,
+    "b": 15,
+    "c": 41,
+    "d": 27,
+    "e": 16,
+    "f": 35,
+    "g": 24,
+    "h": 40,
+    "i": 6,
+    "j": 29,
+    "k": 18,
+    "l": 11,
+    "m": 28,
+    "n": 5,
+    "o": 12,
+    "p": 26,
+    "q": 64,
+    "r": 14,
+    "s": 20,
+    "t": 9,
+    "u": 30,
+    "v": 47,
+    "w": 19,
+    "x": 58,
+    "y": 33,
+    "z": 51,
+    "|": 4,
+    "}": 81,
+    "£": 85,
+    "°": 82,
+    "²": 87,
+    "´": 65,
+    "à": 13,
+    "á": 22,
+    "ç": 79,
+    "è": 37,
+    "é": 36,
+    "ë": 70,
+    "ì": 21,
+    "í": 10,
+    "ï": 89,
+    "ò": 34,
+    "ó": 31,
+    "õ": 77,
+    "ù": 39,
+    "ú": 32,
+    "ü": 71,
+    "ę": 61,
+    "ń": 42,
+    "ǹ": 49,
+    "̀": 23,
+    "́": 25,
+    "̣": 75,
+    "ḿ": 68,
+    "ṛ": 92,
+    "ṣ": 38,
+    "ạ": 93,
+    "ẹ": 17,
+    "ọ": 8,
+    "–": 90,
+    "—": 73,
+    "‘": 76,
+    "’": 69,
+    "”": 60
+  },
+  "yre": {
+    "'": 7,
+    "-": 10,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 15,
+    "c": 29,
+    "d": 16,
+    "e": 11,
+    "f": 31,
+    "g": 32,
+    "h": 34,
+    "i": 9,
+    "j": 33,
+    "k": 18,
+    "l": 12,
+    "m": 20,
+    "n": 5,
+    "o": 24,
+    "p": 27,
+    "r": 22,
+    "s": 28,
+    "t": 17,
+    "u": 25,
+    "v": 30,
+    "w": 21,
+    "y": 19,
+    "z": 23,
+    "|": 4,
+    "ɔ": 14,
+    "ɛ": 8,
+    "ɩ": 13,
+    "ʋ": 26
+  },
+  "yua": {
+    "'": 6,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 13,
+    "d": 28,
+    "e": 7,
+    "f": 35,
+    "g": 34,
+    "h": 23,
+    "i": 12,
+    "j": 16,
+    "k": 21,
+    "l": 8,
+    "m": 17,
+    "n": 14,
+    "o": 10,
+    "p": 25,
+    "q": 27,
+    "r": 33,
+    "s": 22,
+    "t": 11,
+    "u": 9,
+    "v": 36,
+    "w": 29,
+    "x": 19,
+    "y": 18,
+    "z": 26,
+    "|": 4,
+    "á": 20,
+    "é": 24,
+    "í": 32,
+    "ñ": 37,
+    "ó": 31,
+    "ú": 30
+  },
+  "yue-script_traditional": {
+    "\"": 1997,
+    "%": 870,
+    "&": 1972,
+    "+": 1930,
+    ",": 512,
+    "-": 609,
+    ".": 750,
+    "/": 1013,
+    "0": 12,
+    "1": 14,
+    "2": 48,
+    "3": 164,
+    "4": 129,
+    "5": 128,
+    "6": 202,
+    "7": 239,
+    "8": 156,
+    "9": 86,
+    ":": 1060,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 139,
+    "b": 697,
+    "c": 349,
+    "d": 284,
+    "e": 162,
+    "f": 1390,
+    "g": 433,
+    "h": 397,
+    "i": 203,
+    "j": 1472,
+    "k": 809,
+    "l": 287,
+    "m": 399,
+    "n": 201,
+    "o": 177,
+    "p": 396,
+    "q": 1694,
+    "r": 218,
+    "s": 234,
+    "t": 190,
+    "u": 376,
+    "v": 869,
+    "w": 785,
+    "x": 1392,
+    "y": 678,
+    "z": 2262,
+    "·": 296,
+    "í": 1775,
+    "ó": 1776,
+    "õ": 1998,
+    "​": 2154,
+    "–": 1545,
+    "—": 2273,
+    "‧": 356,
+    "─": 1200,
+    "、": 28,
+    "〈": 1796,
+    "〉": 1797,
+    "《": 605,
+    "》": 606,
+    "「": 56,
+    "」": 61,
+    "『": 2127,
+    "』": 2128,
+    "・": 1257,
+    "一": 6,
+    "丁": 1000,
+    "七": 810,
+    "丈": 1860,
+    "三": 249,
+    "上": 27,
+    "下": 87,
+    "不": 23,
+    "且": 222,
+    "世": 131,
+    "丘": 1189,
+    "丟": 1421,
+    "並": 64,
+    "中": 25,
+    "串": 1379,
+    "丹": 891,
+    "主": 133,
+    "乃": 1567,
+    "久": 732,
+    "之": 37,
+    "乎": 554,
+    "乏": 1371,
+    "乘": 893,
+    "九": 829,
+    "也": 45,
+    "乳": 1665,
+    "乾": 1464,
+    "亂": 1256,
+    "了": 22,
+    "予": 1078,
+    "事": 88,
+    "二": 232,
+    "互": 723,
+    "五": 534,
+    "井": 1643,
+    "些": 57,
+    "亞": 79,
+    "亡": 759,
+    "交": 373,
+    "亥": 1983,
+    "亦": 807,
+    "亨": 1501,
+    "享": 871,
+    "京": 848,
+    "亭": 1723,
+    "亮": 1175,
+    "人": 9,
+    "什": 710,
+    "仁": 1787,
+    "仇": 1647,
+    "今": 258,
+    "介": 2146,
+    "仍": 427,
+    "仔": 2287,
+    "他": 19,
+    "仗": 1842,
+    "付": 825,
+    "代": 236,
+    "令": 637,
+    "以": 21,
+    "仰": 890,
+    "仲": 2017,
+    "件": 306,
+    "任": 241,
+    "份": 395,
+    "仿": 1533,
+    "企": 1289,
+    "伊": 455,
+    "伍": 2072,
+    "伏": 1777,
+    "伐": 1693,
+    "休": 1065,
+    "伯": 639,
+    "估": 901,
+    "伴": 1149,
+    "伸": 2047,
+    "伺": 2156,
+    "似": 688,
+    "伽": 1357,
+    "但": 67,
+    "佈": 1641,
+    "位": 170,
+    "低": 508,
+    "住": 331,
+    "佐": 1349,
+    "佔": 1125,
+    "何": 336,
+    "佛": 1137,
+    "作": 96,
+    "你": 120,
+    "佩": 1217,
+    "佳": 780,
+    "併": 2191,
+    "使": 97,
+    "來": 39,
+    "例": 468,
+    "侖": 2211,
+    "供": 279,
+    "依": 692,
+    "侵": 948,
+    "侶": 1861,
+    "侷": 2181,
+    "便": 562,
+    "係": 826,
+    "促": 1181,
+    "俄": 763,
+    "俏": 1807,
+    "俗": 2086,
+    "俘": 1778,
+    "俚": 1542,
+    "保": 267,
+    "信": 466,
+    "修": 419,
+    "俯": 2108,
+    "俱": 1086,
+    "個": 36,
+    "倍": 1139,
+    "們": 32,
+    "倒": 1763,
+    "倖": 1389,
+    "候": 986,
+    "倚": 1830,
+    "借": 1655,
+    "倡": 1837,
+    "倦": 1829,
+    "倫": 793,
+    "值": 1097,
+    "假": 1415,
+    "偉": 1912,
+    "偏": 1288,
+    "做": 374,
+    "停": 856,
+    "健": 583,
+    "側": 1193,
+    "偵": 1956,
+    "偶": 2118,
+    "偷": 1974,
+    "傑": 897,
+    "備": 470,
+    "催": 2139,
+    "傲": 2052,
+    "傳": 302,
+    "債": 2260,
+    "傷": 411,
+    "傾": 1730,
+    "僅": 918,
+    "像": 339,
+    "價": 431,
+    "儀": 1562,
+    "億": 899,
+    "儕": 2075,
+    "儘": 841,
+    "優": 740,
+    "儲": 1388,
+    "允": 1685,
+    "元": 552,
+    "充": 722,
+    "兆": 2237,
+    "先": 393,
+    "光": 317,
+    "克": 143,
+    "兌": 1473,
+    "免": 713,
+    "兒": 782,
+    "兔": 2176,
+    "入": 148,
+    "內": 200,
+    "全": 196,
+    "兩": 318,
+    "八": 1038,
+    "公": 75,
+    "六": 849,
+    "共": 521,
+    "兵": 965,
+    "其": 43,
+    "具": 436,
+    "典": 708,
+    "冊": 1577,
+    "再": 424,
+    "冒": 1347,
+    "冠": 925,
+    "冬": 824,
+    "冰": 402,
+    "冷": 1083,
+    "准": 944,
+    "凌": 1454,
+    "凡": 1441,
+    "凱": 1325,
+    "出": 65,
+    "函": 2152,
+    "刀": 1996,
+    "分": 70,
+    "切": 771,
+    "刊": 1905,
+    "刑": 1451,
+    "划": 1662,
+    "列": 737,
+    "初": 650,
+    "判": 868,
+    "別": 409,
+    "利": 101,
+    "刪": 1376,
+    "到": 29,
+    "制": 400,
+    "刷": 2066,
+    "券": 1622,
+    "刺": 1201,
+    "刻": 1159,
+    "則": 371,
+    "削": 1381,
+    "前": 81,
+    "剖": 1929,
+    "剛": 1397,
+    "剝": 1785,
+    "剩": 1232,
+    "剪": 1849,
+    "副": 1609,
+    "割": 1915,
+    "創": 530,
+    "劃": 953,
+    "劇": 983,
+    "劍": 1369,
+    "劑": 2197,
+    "力": 141,
+    "功": 596,
+    "加": 83,
+    "劣": 1850,
+    "助": 367,
+    "努": 767,
+    "劫": 2037,
+    "勃": 1901,
+    "勇": 2238,
+    "勒": 819,
+    "動": 35,
+    "勘": 2178,
+    "務": 347,
+    "勝": 929,
+    "勞": 1291,
+    "勢": 1179,
+    "勤": 1550,
+    "勵": 1924,
+    "勸": 1691,
+    "包": 435,
+    "匈": 1406,
+    "化": 192,
+    "北": 277,
+    "匙": 2046,
+    "匯": 1521,
+    "匹": 931,
+    "匿": 1736,
+    "區": 147,
+    "十": 282,
+    "千": 502,
+    "升": 994,
+    "午": 1304,
+    "卉": 2080,
+    "半": 812,
+    "卓": 1751,
+    "協": 543,
+    "南": 260,
+    "博": 806,
+    "占": 1762,
+    "卡": 379,
+    "印": 364,
+    "危": 649,
+    "即": 460,
+    "卵": 1652,
+    "卷": 2285,
+    "卸": 1244,
+    "卻": 686,
+    "厄": 1351,
+    "厚": 1108,
+    "原": 181,
+    "厭": 1564,
+    "厲": 1894,
+    "去": 323,
+    "參": 513,
+    "又": 724,
+    "及": 99,
+    "友": 910,
+    "反": 426,
+    "取": 322,
+    "受": 121,
+    "叢": 1469,
+    "口": 360,
+    "古": 495,
+    "句": 1238,
+    "另": 580,
+    "只": 255,
+    "叫": 1544,
+    "可": 26,
+    "台": 919,
+    "史": 289,
+    "右": 837,
+    "司": 311,
+    "吃": 1064,
+    "各": 388,
+    "合": 251,
+    "吉": 1006,
+    "同": 108,
+    "名": 82,
+    "后": 1539,
+    "吐": 1782,
+    "向": 478,
+    "君": 2004,
+    "否": 610,
+    "吧": 1240,
+    "含": 614,
+    "吳": 1978,
+    "吸": 1034,
+    "吹": 1964,
+    "吼": 1933,
+    "���": 1678,
+    "呆": 1918,
+    "呈": 1185,
+    "告": 429,
+    "周": 600,
+    "味": 987,
+    "呼": 1323,
+    "命": 533,
+    "和": 15,
+    "咎": 1958,
+    "咒": 2142,
+    "咖": 1088,
+    "咬": 1659,
+    "品": 398,
+    "哇": 1863,
+    "哈": 902,
+    "員": 179,
+    "哥": 645,
+    "哩": 1908,
+    "哪": 1633,
+    "哭": 2000,
+    "哲": 1976,
+    "哺": 2112,
+    "唐": 1931,
+    "售": 1230,
+    "唯": 860,
+    "唱": 1332,
+    "商": 581,
+    "問": 442,
+    "啟": 845,
+    "啡": 1195,
+    "喀": 2115,
+    "善": 1135,
+    "喉": 2053,
+    "喙": 2241,
+    "喜": 936,
+    "喝": 1981,
+    "喪": 1939,
+    "喬": 1128,
+    "單": 300,
+    "嘆": 1720,
+    "嘔": 1781,
+    "嘗": 1157,
+    "嘯": 884,
+    "嘲": 2010,
+    "器": 503,
+    "噱": 2147,
+    "噴": 1101,
+    "嚇": 1637,
+    "嚐": 2015,
+    "嚨": 2054,
+    "嚴": 632,
+    "囊": 2065,
+    "囚": 1233,
+    "四": 417,
+    "回": 293,
+    "因": 73,
+    "困": 1058,
+    "固": 1620,
+    "圈": 1300,
+    "國": 11,
+    "圍": 719,
+    "園": 588,
+    "圓": 1719,
+    "圖": 352,
+    "團": 481,
+    "土": 430,
+    "在": 5,
+    "圭": 1372,
+    "地": 16,
+    "圾": 2133,
+    "址": 1106,
+    "均": 1428,
+    "坍": 2235,
+    "坎": 2022,
+    "坐": 1190,
+    "坑": 1591,
+    "坡": 950,
+    "坦": 532,
+    "垃": 2132,
+    "型": 230,
+    "埃": 721,
+    "埋": 2189,
+    "城": 295,
+    "埔": 1365,
+    "域": 628,
+    "執": 1115,
+    "培": 1498,
+    "基": 220,
+    "堂": 1360,
+    "堅": 1321,
+    "堆": 1619,
+    "堡": 370,
+    "堤": 2092,
+    "堪": 1679,
+    "報": 265,
+    "場": 130,
+    "堵": 1411,
+    "塊": 1344,
+    "塌": 1587,
+    "塑": 1721,
+    "塔": 585,
+    "塗": 1596,
+    "塚": 2082,
+    "塞": 694,
+    "填": 1416,
+    "塵": 1885,
+    "境": 549,
+    "墓": 1258,
+    "墜": 1618,
+    "增": 684,
+    "墨": 1938,
+    "墮": 1556,
+    "墳": 1739,
+    "壁": 1307,
+    "壇": 2096,
+    "壓": 792,
+    "壘": 1399,
+    "壞": 839,
+    "壤": 2242,
+    "士": 358,
+    "壯": 1952,
+    "夏": 574,
+    "夕": 1826,
+    "外": 157,
+    "多": 30,
+    "夜": 1066,
+    "夠": 654,
+    "夢": 1168,
+    "夥": 1928,
+    "大": 17,
+    "天": 154,
+    "太": 229,
+    "夫": 570,
+    "央": 1171,
+    "失": 484,
+    "夾": 1770,
+    "奇": 752,
+    "奈": 1973,
+    "奉": 1907,
+    "奏": 2202,
+    "契": 1435,
+    "奔": 1563,
+    "套": 1352,
+    "奠": 2216,
+    "奢": 1847,
+    "奧": 270,
+    "奪": 1255,
+    "奮": 1740,
+    "女": 704,
+    "奶": 1442,
+    "她": 567,
+    "好": 384,
+    "如": 135,
+    "妙": 2217,
+    "妥": 1724,
+    "妨": 1395,
+    "妮": 1505,
+    "妹": 2040,
+    "妻": 1338,
+    "姆": 928,
+    "姊": 2039,
+    "始": 321,
+    "姓": 1742,
+    "委": 651,
+    "姻": 1500,
+    "姿": 1110,
+    "威": 452,
+    "娛": 2158,
+    "娜": 1882,
+    "婚": 1081,
+    "婦": 1272,
+    "婪": 1919,
+    "媒": 954,
+    "媲": 1756,
+    "嫌": 943,
+    "嬰": 2136,
+    "子": 95,
+    "孔": 1875,
+    "字": 405,
+    "存": 423,
+    "季": 351,
+    "孤": 1859,
+    "孩": 670,
+    "孫": 2129,
+    "學": 71,
+    "它": 146,
+    "宅": 1422,
+    "宇": 1283,
+    "守": 859,
+    "安": 355,
+    "完": 518,
+    "宗": 935,
+    "官": 496,
+    "宙": 1373,
+    "定": 114,
+    "宜": 1015,
+    "客": 326,
+    "宣": 621,
+    "室": 977,
+    "宮": 2138,
+    "宰": 1487,
+    "害": 457,
+    "宴": 1611,
+    "家": 53,
+    "容": 617,
+    "宿": 823,
+    "寂": 1813,
+    "寄": 903,
+    "密": 742,
+    "寇": 1780,
+    "富": 624,
+    "寒": 1725,
+    "寓": 1302,
+    "察": 416,
+    "寢": 1848,
+    "實": 310,
+    "寧": 1718,
+    "寨": 1366,
+    "審": 818,
+    "寫": 730,
+    "寬": 1143,
+    "寶": 1249,
+    "寸": 1481,
+    "寺": 1354,
+    "封": 930,
+    "射": 636,
+    "將": 85,
+    "專": 696,
+    "尊": 1575,
+    "尋": 795,
+    "對": 78,
+    "導": 335,
+    "小": 151,
+    "少": 244,
+    "尖": 1339,
+    "尚": 1018,
+    "尤": 985,
+    "就": 89,
+    "尺": 976,
+    "尼": 309,
+    "尾": 1517,
+    "尿": 1095,
+    "局": 474,
+    "居": 432,
+    "屆": 1383,
+    "屈": 1638,
+    "屋": 1274,
+    "屏": 2083,
+    "展": 540,
+    "屠": 1624,
+    "層": 836,
+    "屬": 523,
+    "山": 312,
+    "岡": 1393,
+    "岩": 851,
+    "岳": 1590,
+    "岸": 1061,
+    "峭": 1306,
+    "峰": 1335,
+    "島": 304,
+    "峻": 2203,
+    "峽": 1212,
+    "崇": 1818,
+    "崎": 1312,
+    "崔": 2090,
+    "崖": 2110,
+    "崗": 1858,
+    "崙": 2186,
+    "崩": 1506,
+    "嶄": 2190,
+    "嶇": 1704,
+    "嶺": 2196,
+    "嶼": 1025,
+    "川": 1152,
+    "州": 332,
+    "巡": 1674,
+    "巢": 2062,
+    "工": 275,
+    "左": 907,
+    "巧": 1396,
+    "巨": 1162,
+    "差": 699,
+    "己": 343,
+    "已": 161,
+    "巴": 221,
+    "巽": 2240,
+    "巾": 2101,
+    "市": 315,
+    "布": 301,
+    "希": 492,
+    "帕": 783,
+    "帝": 1063,
+    "師": 563,
+    "席": 756,
+    "帳": 2011,
+    "帶": 243,
+    "常": 50,
+    "幅": 959,
+    "幕": 999,
+    "幟": 1967,
+    "幣": 1220,
+    "幫": 857,
+    "平": 329,
+    "年": 31,
+    "幸": 1525,
+    "幹": 1686,
+    "幻": 2269,
+    "幼": 2135,
+    "幾": 273,
+    "床": 1243,
+    "序": 667,
+    "底": 707,
+    "店": 674,
+    "府": 439,
+    "度": 102,
+    "座": 338,
+    "庫": 864,
+    "庭": 607,
+    "康": 791,
+    "廁": 2032,
+    "廂": 1424,
+    "廉": 1661,
+    "廊": 2008,
+    "廟": 676,
+    "廠": 1541,
+    "廢": 2157,
+    "廣": 598,
+    "廳": 790,
+    "延": 969,
+    "廷": 1343,
+    "建": 271,
+    "式": 125,
+    "弓": 726,
+    "引": 613,
+    "弗": 998,
+    "弧": 2171,
+    "弱": 1051,
+    "張": 989,
+    "強": 425,
+    "彈": 832,
+    "彌": 1803,
+    "彎": 1457,
+    "彗": 2023,
+    "彙": 1402,
+    "形": 378,
+    "彩": 1499,
+    "彭": 2058,
+    "彰": 1708,
+    "影": 208,
+    "役": 1531,
+    "彼": 952,
+    "往": 348,
+    "征": 962,
+    "待": 619,
+    "很": 216,
+    "律": 553,
+    "後": 49,
+    "徑": 1285,
+    "徒": 1010,
+    "得": 76,
+    "徙": 1182,
+    "從": 118,
+    "復": 878,
+    "循": 1755,
+    "微": 577,
+    "徵": 1049,
+    "德": 140,
+    "徹": 1414,
+    "心": 212,
+    "必": 237,
+    "志": 1156,
+    "忘": 1692,
+    "快": 556,
+    "念": 753,
+    "忽": 2016,
+    "怎": 1921,
+    "怒": 1735,
+    "怕": 1173,
+    "怖": 1874,
+    "思": 698,
+    "急": 1074,
+    "性": 205,
+    "怨": 1975,
+    "怪": 1268,
+    "恆": 1466,
+    "恐": 681,
+    "恢": 1294,
+    "恩": 905,
+    "息": 653,
+    "悅": 1404,
+    "悉": 1407,
+    "悖": 1700,
+    "悟": 1511,
+    "悠": 1910,
+    "患": 911,
+    "您": 268,
+    "情": 408,
+    "惜": 2165,
+    "惠": 1194,
+    "惡": 990,
+    "惱": 2205,
+    "想": 461,
+    "惹": 2224,
+    "意": 185,
+    "愛": 558,
+    "感": 403,
+    "慈": 2168,
+    "態": 488,
+    "慌": 2074,
+    "慎": 1440,
+    "慘": 1684,
+    "慢": 1358,
+    "慣": 1536,
+    "慧": 1713,
+    "慮": 1080,
+    "慶": 1043,
+    "憂": 1290,
+    "憑": 2173,
+    "憤": 1734,
+    "憲": 858,
+    "憶": 1219,
+    "憾": 1845,
+    "應": 209,
+    "懲": 2175,
+    "懶": 2104,
+    "懷": 1537,
+    "懸": 1705,
+    "懼": 1213,
+    "戈": 1385,
+    "成": 34,
+    "我": 115,
+    "戒": 2103,
+    "或": 42,
+    "戛": 1460,
+    "截": 1341,
+    "戰": 303,
+    "戲": 941,
+    "戴": 1259,
+    "戶": 803,
+    "房": 626,
+    "所": 66,
+    "手": 176,
+    "才": 334,
+    "扎": 1749,
+    "打": 328,
+    "扔": 1994,
+    "托": 992,
+    "扣": 2126,
+    "扮": 2055,
+    "扯": 1573,
+    "扶": 1629,
+    "批": 671,
+    "找": 634,
+    "承": 738,
+    "技": 593,
+    "抄": 2148,
+    "把": 744,
+    "抑": 1753,
+    "投": 640,
+    "抗": 546,
+    "折": 1523,
+    "抬": 2274,
+    "抱": 1906,
+    "抵": 815,
+    "抹": 2272,
+    "抽": 2027,
+    "拉": 163,
+    "拋": 1949,
+    "拍": 1301,
+    "拒": 997,
+    "拔": 1446,
+    "拖": 1027,
+    "拘": 1295,
+    "招": 1555,
+    "拜": 1270,
+    "括": 695,
+    "拯": 2122,
+    "拳": 1795,
+    "拼": 1608,
+    "拿": 693,
+    "持": 235,
+    "指": 280,
+    "按": 2184,
+    "挑": 1380,
+    "挖": 2078,
+    "挪": 1337,
+    "挫": 2172,
+    "振": 1408,
+    "挺": 1548,
+    "捕": 717,
+    "捲": 972,
+    "捷": 2292,
+    "掃": 1543,
+    "授": 1012,
+    "掉": 885,
+    "掌": 1582,
+    "排": 575,
+    "掠": 2254,
+    "採": 1028,
+    "探": 964,
+    "接": 204,
+    "控": 420,
+    "推": 566,
+    "掩": 1616,
+    "措": 1628,
+    "描": 1234,
+    "提": 138,
+    "插": 1160,
+    "揚": 1672,
+    "換": 876,
+    "握": 1299,
+    "揣": 1878,
+    "揭": 1275,
+    "揮": 1036,
+    "援": 1047,
+    "損": 720,
+    "搖": 2174,
+    "搜": 1077,
+    "搞": 1508,
+    "搥": 1793,
+    "搬": 1471,
+    "搭": 1071,
+    "摔": 1492,
+    "摘": 1791,
+    "摧": 1917,
+    "摩": 736,
+    "摺": 828,
+    "撐": 1988,
+    "撒": 2188,
+    "撞": 1123,
+    "撤": 1271,
+    "撥": 1331,
+    "撫": 1668,
+    "播": 779,
+    "撰": 1604,
+    "撼": 2295,
+    "擁": 833,
+    "擅": 1549,
+    "擇": 834,
+    "擊": 324,
+    "擋": 1600,
+    "操": 1263,
+    "擎": 1632,
+    "擔": 1007,
+    "據": 256,
+    "擠": 1558,
+    "擦": 1936,
+    "擬": 1330,
+    "擲": 1677,
+    "擴": 2067,
+    "擺": 1613,
+    "擾": 1731,
+    "攀": 1079,
+    "攔": 2257,
+    "攜": 872,
+    "攝": 746,
+    "支": 456,
+    "收": 571,
+    "改": 363,
+    "攻": 498,
+    "放": 350,
+    "政": 298,
+    "故": 519,
+    "效": 689,
+    "敏": 1367,
+    "救": 770,
+    "敗": 776,
+    "教": 254,
+    "敞": 2263,
+    "散": 1278,
+    "敦": 1180,
+    "敲": 1606,
+    "整": 644,
+    "敵": 1439,
+    "數": 122,
+    "文": 126,
+    "斗": 1962,
+    "料": 620,
+    "斜": 1648,
+    "斤": 1816,
+    "斯": 55,
+    "新": 158,
+    "斷": 597,
+    "方": 47,
+    "於": 38,
+    "施": 932,
+    "旁": 1362,
+    "旅": 150,
+    "旋": 1316,
+    "族": 945,
+    "旗": 1318,
+    "既": 2255,
+    "日": 113,
+    "旦": 1426,
+    "旨": 1683,
+    "早": 608,
+    "旱": 2206,
+    "旺": 1758,
+    "昂": 1520,
+    "昆": 1656,
+    "明": 197,
+    "易": 464,
+    "昔": 1948,
+    "星": 327,
+    "映": 1794,
+    "春": 1610,
+    "昨": 1726,
+    "昭": 1707,
+    "是": 7,
+    "時": 18,
+    "晉": 2194,
+    "晚": 664,
+    "晝": 1943,
+    "晤": 2159,
+    "晨": 1315,
+    "普": 550,
+    "景": 627,
+    "晶": 2225,
+    "智": 1188,
+    "暈": 1880,
+    "暑": 1886,
+    "暖": 1824,
+    "暗": 1197,
+    "暫": 981,
+    "暱": 1737,
+    "暴": 794,
+    "暹": 2266,
+    "曆": 2167,
+    "曉": 2213,
+    "曝": 1594,
+    "曲": 1482,
+    "更": 92,
+    "書": 873,
+    "曼": 926,
+    "曾": 447,
+    "替": 1817,
+    "最": 68,
+    "會": 13,
+    "月": 106,
+    "有": 8,
+    "朋": 1769,
+    "服": 385,
+    "朗": 1581,
+    "望": 537,
+    "朝": 712,
+    "期": 193,
+    "木": 773,
+    "未": 337,
+    "末": 1218,
+    "本": 136,
+    "札": 1586,
+    "朱": 2210,
+    "杉": 1745,
+    "李": 1617,
+    "材": 1515,
+    "村": 1774,
+    "杜": 1574,
+    "束": 658,
+    "杭": 1767,
+    "東": 362,
+    "松": 1790,
+    "板": 886,
+    "析": 2031,
+    "林": 413,
+    "枚": 1682,
+    "果": 187,
+    "枝": 1660,
+    "架": 1154,
+    "柏": 1114,
+    "某": 357,
+    "染": 808,
+    "柔": 1223,
+    "查": 263,
+    "柬": 1364,
+    "柯": 2003,
+    "柱": 1833,
+    "栓": 2076,
+    "校": 1055,
+    "株": 1852,
+    "核": 601,
+    "根": 404,
+    "格": 291,
+    "框": 2182,
+    "案": 375,
+    "桌": 2282,
+    "桑": 1462,
+    "桶": 1266,
+    "桿": 1547,
+    "梅": 1427,
+    "條": 643,
+    "梠": 1834,
+    "梨": 2281,
+    "梭": 1950,
+    "梯": 1927,
+    "梵": 916,
+    "棄": 1519,
+    "棍": 1993,
+    "棒": 1495,
+    "棕": 1386,
+    "棘": 2234,
+    "棚": 2249,
+    "棟": 1623,
+    "森": 618,
+    "棺": 1940,
+    "椅": 1522,
+    "植": 1042,
+    "椰": 1909,
+    "楚": 1711,
+    "業": 269,
+    "極": 406,
+    "概": 1303,
+    "榜": 1423,
+    "榨": 1946,
+    "榮": 1478,
+    "榻": 1650,
+    "構": 602,
+    "槍": 1109,
+    "槽": 1342,
+    "樂": 545,
+    "樓": 1329,
+    "標": 544,
+    "模": 747,
+    "樣": 238,
+    "樹": 701,
+    "橄": 1286,
+    "橇": 1968,
+    "橋": 1098,
+    "機": 159,
+    "橫": 1121,
+    "檢": 680,
+    "檯": 1589,
+    "櫃": 2035,
+    "櫚": 1980,
+    "櫻": 2277,
+    "權": 499,
+    "欖": 1287,
+    "次": 206,
+    "欣": 2069,
+    "欲": 1987,
+    "欺": 2073,
+    "款": 1033,
+    "歌": 1070,
+    "歎": 2095,
+    "歐": 346,
+    "歡": 900,
+    "止": 647,
+    "正": 191,
+    "此": 62,
+    "步": 564,
+    "武": 1017,
+    "歧": 1759,
+    "歲": 817,
+    "歷": 381,
+    "歸": 1453,
+    "死": 539,
+    "殊": 1296,
+    "殖": 1127,
+    "殘": 1914,
+    "段": 797,
+    "殺": 1026,
+    "殼": 1561,
+    "殿": 917,
+    "毀": 1089,
+    "母": 590,
+    "每": 262,
+    "毒": 631,
+    "比": 90,
+    "毛": 1005,
+    "毫": 1340,
+    "氏": 1132,
+    "民": 188,
+    "氛": 1839,
+    "氣": 307,
+    "氧": 1765,
+    "氫": 1764,
+    "氮": 2005,
+    "氰": 2026,
+    "水": 134,
+    "永": 949,
+    "汀": 1293,
+    "汁": 1947,
+    "求": 477,
+    "汐": 2248,
+    "汗": 1004,
+    "汙": 1925,
+    "池": 1664,
+    "污": 2028,
+    "決": 313,
+    "汽": 787,
+    "沃": 1540,
+    "沉": 1916,
+    "沒": 226,
+    "沙": 727,
+    "沫": 2195,
+    "河": 507,
+    "油": 517,
+    "治": 576,
+    "沼": 2030,
+    "沿": 846,
+    "況": 448,
+    "泊": 1766,
+    "法": 44,
+    "泛": 1085,
+    "泡": 960,
+    "波": 443,
+    "泣": 2001,
+    "泥": 1681,
+    "注": 772,
+    "泰": 912,
+    "泳": 1187,
+    "泵": 1690,
+    "洋": 525,
+    "洗": 1645,
+    "洛": 489,
+    "洞": 774,
+    "洩": 1117,
+    "洪": 1578,
+    "洲": 199,
+    "活": 175,
+    "洽": 1879,
+    "派": 1141,
+    "流": 278,
+    "浩": 1671,
+    "浪": 1324,
+    "浮": 1177,
+    "浴": 2251,
+    "海": 168,
+    "涅": 1588,
+    "消": 450,
+    "涉": 1119,
+    "涯": 1673,
+    "液": 1009,
+    "涵": 1458,
+    "涸": 2036,
+    "涼": 2201,
+    "淆": 1890,
+    "淚": 2114,
+    "淨": 1094,
+    "深": 842,
+    "混": 1497,
+    "淹": 2091,
+    "清": 734,
+    "減": 867,
+    "渡": 2296,
+    "測": 494,
+    "港": 1113,
+    "渴": 1971,
+    "游": 768,
+    "渺": 2230,
+    "湖": 862,
+    "湯": 1391,
+    "源": 462,
+    "準": 386,
+    "溜": 1308,
+    "溝": 1954,
+    "溢": 1904,
+    "溪": 1592,
+    "溫": 799,
+    "溼": 1926,
+    "滅": 1876,
+    "滋": 1514,
+    "滑": 483,
+    "滕": 2215,
+    "滾": 1728,
+    "滿": 612,
+    "漂": 2087,
+    "漆": 1768,
+    "漏": 1431,
+    "演": 506,
+    "漠": 2089,
+    "漢": 1072,
+    "漫": 1103,
+    "漸": 1461,
+    "漿": 1467,
+    "潔": 2160,
+    "潛": 589,
+    "潤": 1475,
+    "潦": 2107,
+    "潮": 1899,
+    "澀": 1867,
+    "澡": 2252,
+    "澤": 1741,
+    "澳": 652,
+    "激": 921,
+    "濃": 1142,
+    "濘": 1865,
+    "濟": 757,
+    "濤": 1922,
+    "濫": 2009,
+    "濾": 2185,
+    "瀆": 2278,
+    "瀉": 2143,
+    "瀑": 1565,
+    "灘": 1363,
+    "灣": 875,
+    "火": 526,
+    "灼": 2212,
+    "災": 1091,
+    "炎": 1518,
+    "炭": 1892,
+    "炸": 741,
+    "為": 10,
+    "烈": 843,
+    "烏": 1199,
+    "烘": 2221,
+    "烤": 2222,
+    "烹": 1211,
+    "無": 207,
+    "焦": 852,
+    "然": 142,
+    "煙": 1806,
+    "煤": 1891,
+    "照": 623,
+    "煩": 1405,
+    "煮": 1273,
+    "熊": 1297,
+    "熟": 1163,
+    "熨": 1961,
+    "熱": 407,
+    "燃": 1676,
+    "燈": 811,
+    "燒": 1489,
+    "營": 641,
+    "爆": 687,
+    "爐": 1688,
+    "爪": 946,
+    "爬": 1911,
+    "爭": 480,
+    "爵": 1429,
+    "父": 1333,
+    "爾": 127,
+    "牆": 1599,
+    "片": 401,
+    "版": 865,
+    "牌": 1284,
+    "牙": 633,
+    "牛": 993,
+    "牠": 630,
+    "牢": 1831,
+    "牧": 1252,
+    "物": 52,
+    "特": 91,
+    "牽": 1630,
+    "犒": 2068,
+    "犬": 1887,
+    "犯": 889,
+    "狀": 383,
+    "狂": 1819,
+    "狄": 2288,
+    "狗": 2111,
+    "狩": 2019,
+    "狼": 1788,
+    "猁": 2049,
+    "猛": 840,
+    "猜": 1808,
+    "猞": 2048,
+    "猩": 1480,
+    "猴": 2218,
+    "猶": 1394,
+    "猿": 2294,
+    "獄": 1111,
+    "獅": 1228,
+    "獎": 861,
+    "獨": 559,
+    "獲": 285,
+    "獵": 850,
+    "獸": 2012,
+    "獺": 1951,
+    "獻": 1403,
+    "玆": 1984,
+    "率": 715,
+    "王": 444,
+    "玩": 1443,
+    "玻": 1215,
+    "珀": 1970,
+    "珊": 1697,
+    "珍": 1572,
+    "班": 476,
+    "現": 74,
+    "球": 80,
+    "理": 103,
+    "琴": 1207,
+    "瑞": 584,
+    "瑟": 1221,
+    "瑪": 1151,
+    "璃": 1470,
+    "環": 578,
+    "瓜": 1041,
+    "瓣": 1965,
+    "瓦": 1019,
+    "瓶": 1452,
+    "甘": 1870,
+    "甚": 473,
+    "生": 41,
+    "產": 266,
+    "甦": 2204,
+    "用": 46,
+    "田": 2284,
+    "由": 144,
+    "甲": 1535,
+    "申": 1062,
+    "男": 906,
+    "界": 240,
+    "留": 604,
+    "畜": 2283,
+    "畢": 1126,
+    "略": 888,
+    "畫": 560,
+    "異": 728,
+    "當": 100,
+    "疊": 1485,
+    "疏": 2041,
+    "疑": 1092,
+    "疫": 1262,
+    "疲": 1646,
+    "疼": 1361,
+    "疽": 2120,
+    "疾": 924,
+    "病": 389,
+    "症": 844,
+    "痛": 1118,
+    "瘓": 1942,
+    "瘤": 1800,
+    "瘧": 2071,
+    "療": 1242,
+    "癌": 1328,
+    "癒": 2125,
+    "癱": 1941,
+    "登": 565,
+    "發": 51,
+    "白": 625,
+    "百": 700,
+    "的": 4,
+    "皆": 934,
+    "皇": 894,
+    "皮": 1172,
+    "盃": 2021,
+    "盈": 1868,
+    "益": 1448,
+    "盎": 1717,
+    "盒": 2239,
+    "盔": 1889,
+    "盛": 1204,
+    "盜": 1465,
+    "盟": 657,
+    "盡": 963,
+    "監": 735,
+    "盤": 1783,
+    "盧": 933,
+    "盪": 2233,
+    "目": 211,
+    "直": 257,
+    "相": 160,
+    "盾": 1761,
+    "省": 915,
+    "看": 290,
+    "真": 660,
+    "眠": 1400,
+    "眼": 980,
+    "眾": 511,
+    "睛": 1801,
+    "睡": 1227,
+    "督": 1148,
+    "睹": 1873,
+    "瞭": 1516,
+    "瞰": 2109,
+    "矛": 1760,
+    "知": 248,
+    "矩": 1898,
+    "短": 683,
+    "矯": 2271,
+    "石": 561,
+    "研": 557,
+    "破": 838,
+    "砸": 2123,
+    "硬": 1186,
+    "碉": 2018,
+    "碎": 1320,
+    "碟": 1593,
+    "碩": 1382,
+    "碰": 1368,
+    "碳": 1893,
+    "確": 219,
+    "碼": 703,
+    "磁": 1504,
+    "磅": 1401,
+    "磯": 1746,
+    "礁": 2033,
+    "礎": 1583,
+    "礙": 1073,
+    "示": 155,
+    "社": 591,
+    "祕": 1282,
+    "祖": 1568,
+    "祝": 1192,
+    "神": 372,
+    "票": 314,
+    "祭": 2169,
+    "禁": 1021,
+    "禍": 1265,
+    "福": 1107,
+    "禦": 2253,
+    "禮": 1002,
+    "禽": 1438,
+    "秀": 1434,
+    "私": 909,
+    "科": 264,
+    "秒": 1206,
+    "秘": 1636,
+    "租": 1348,
+    "移": 509,
+    "稀": 1298,
+    "稅": 880,
+    "程": 184,
+    "稍": 1100,
+    "種": 72,
+    "稱": 153,
+    "積": 940,
+    "穩": 1280,
+    "穫": 1786,
+    "穴": 1752,
+    "究": 573,
+    "空": 194,
+    "穿": 1131,
+    "突": 1045,
+    "窗": 1011,
+    "窟": 1979,
+    "窩": 2094,
+    "窮": 1675,
+    "窯": 1459,
+    "竄": 2286,
+    "竊": 1584,
+    "立": 250,
+    "站": 325,
+    "竟": 2245,
+    "童": 958,
+    "端": 1068,
+    "競": 805,
+    "笑": 1695,
+    "符": 814,
+    "第": 169,
+    "筆": 1001,
+    "等": 223,
+    "筒": 1621,
+    "答": 1580,
+    "策": 1425,
+    "算": 827,
+    "管": 344,
+    "箭": 1888,
+    "箱": 1203,
+    "節": 391,
+    "範": 974,
+    "築": 789,
+    "簡": 706,
+    "簽": 725,
+    "籍": 1959,
+    "籤": 2276,
+    "籲": 1844,
+    "米": 662,
+    "粉": 1710,
+    "粒": 1198,
+    "粗": 1715,
+    "精": 762,
+    "糕": 1468,
+    "糖": 1264,
+    "糟": 1214,
+    "��": 449,
+    "紀": 361,
+    "約": 259,
+    "紅": 1136,
+    "紋": 2161,
+    "納": 531,
+    "紐": 555,
+    "紓": 1843,
+    "純": 1635,
+    "紙": 988,
+    "級": 524,
+    "紛": 1309,
+    "素": 898,
+    "索": 830,
+    "紮": 2088,
+    "累": 1488,
+    "細": 501,
+    "終": 729,
+    "組": 246,
+    "結": 272,
+    "絕": 1050,
+    "絡": 1216,
+    "給": 345,
+    "統": 305,
+    "絲": 1046,
+    "經": 98,
+    "綜": 2056,
+    "綠": 1566,
+    "綢": 2034,
+    "維": 276,
+    "綱": 2081,
+    "網": 288,
+    "綽": 2060,
+    "綿": 1900,
+    "緊": 1035,
+    "線": 394,
+    "緣": 1147,
+    "編": 820,
+    "緩": 1966,
+    "練": 982,
+    "緹": 1985,
+    "緻": 2155,
+    "縣": 2193,
+    "縫": 1387,
+    "縮": 1557,
+    "縱": 1748,
+    "總": 528,
+    "績": 1236,
+    "繁": 1644,
+    "織": 542,
+    "繞": 1069,
+    "繪": 1709,
+    "繫": 2289,
+    "繳": 2259,
+    "繼": 788,
+    "纂": 2093,
+    "續": 412,
+    "纏": 1792,
+    "纖": 1960,
+    "纜": 1493,
+    "缺": 709,
+    "罕": 2187,
+    "罪": 1090,
+    "置": 672,
+    "罰": 1832,
+    "署": 1140,
+    "罹": 1750,
+    "羅": 210,
+    "羊": 1436,
+    "美": 117,
+    "羚": 2177,
+    "群": 438,
+    "義": 365,
+    "羽": 1689,
+    "翅": 1836,
+    "習": 569,
+    "翔": 2079,
+    "翰": 920,
+    "翻": 831,
+    "耀": 2270,
+    "老": 486,
+    "考": 822,
+    "者": 112,
+    "而": 63,
+    "耐": 1410,
+    "耕": 2258,
+    "耗": 1626,
+    "耳": 802,
+    "耶": 1267,
+    "聆": 2070,
+    "聖": 493,
+    "聚": 1576,
+    "聞": 679,
+    "聯": 252,
+    "聰": 1702,
+    "聲": 482,
+    "聳": 2117,
+    "職": 1370,
+    "聽": 1032,
+    "肚": 2140,
+    "肝": 1432,
+    "肢": 2119,
+    "肩": 1534,
+    "肯": 1170,
+    "育": 568,
+    "肺": 1420,
+    "背": 1124,
+    "胎": 1444,
+    "胞": 733,
+    "胡": 1355,
+    "胺": 1528,
+    "能": 24,
+    "脅": 853,
+    "脈": 881,
+    "脫": 967,
+    "脹": 1375,
+    "腦": 572,
+    "腫": 1799,
+    "腰": 1884,
+    "腳": 821,
+    "腸": 2164,
+    "腹": 1716,
+    "膀": 1507,
+    "膜": 1798,
+    "膝": 2192,
+    "膨": 1374,
+    "膽": 1732,
+    "臂": 2149,
+    "臉": 1449,
+    "臘": 956,
+    "臟": 1327,
+    "臣": 2151,
+    "臥": 1864,
+    "臨": 1112,
+    "自": 69,
+    "臭": 2163,
+    "至": 166,
+    "致": 440,
+    "臺": 1093,
+    "與": 54,
+    "興": 668,
+    "舉": 366,
+    "舊": 913,
+    "舍": 2256,
+    "舒": 1245,
+    "舞": 1251,
+    "航": 392,
+    "般": 594,
+    "船": 463,
+    "艇": 743,
+    "艘": 1419,
+    "艙": 739,
+    "艦": 769,
+    "良": 1020,
+    "艱": 1866,
+    "色": 754,
+    "艾": 804,
+    "芬": 896,
+    "芯": 1937,
+    "花": 485,
+    "苗": 1559,
+    "若": 616,
+    "苦": 2099,
+    "英": 149,
+    "范": 1670,
+    "茨": 1823,
+    "茫": 2231,
+    "茱": 1877,
+    "茲": 1353,
+    "茶": 1524,
+    "草": 1144,
+    "荒": 1474,
+    "荷": 883,
+    "莉": 1167,
+    "莊": 2042,
+    "莎": 1527,
+    "莫": 1087,
+    "菊": 1871,
+    "菌": 1030,
+    "菜": 1277,
+    "華": 516,
+    "菲": 1222,
+    "萊": 970,
+    "萬": 445,
+    "落": 421,
+    "葉": 1133,
+    "著": 124,
+    "葛": 1209,
+    "董": 1513,
+    "葬": 1248,
+    "蒂": 922,
+    "蒙": 847,
+    "蒸": 1992,
+    "蓄": 1986,
+    "蓋": 690,
+    "蔡": 2063,
+    "蔥": 1412,
+    "蔽": 2236,
+    "蕪": 1784,
+    "蕾": 2153,
+    "薄": 1054,
+    "薦": 1840,
+    "薩": 603,
+    "薪": 1895,
+    "藉": 1176,
+    "藍": 2199,
+    "藏": 1202,
+    "藝": 677,
+    "藤": 1476,
+    "藥": 1729,
+    "藻": 1538,
+    "蘇": 514,
+    "蘋": 1484,
+    "蘭": 183,
+    "虎": 1450,
+    "虐": 1322,
+    "虔": 1957,
+    "處": 333,
+    "虛": 1191,
+    "虜": 1779,
+    "號": 308,
+    "蚊": 1649,
+    "蛇": 2219,
+    "蛋": 2024,
+    "蜃": 1853,
+    "蝕": 2223,
+    "融": 1082,
+    "螞": 2061,
+    "螢": 1701,
+    "螺": 1991,
+    "蟲": 1053,
+    "蟻": 1827,
+    "血": 1039,
+    "行": 33,
+    "術": 515,
+    "街": 1048,
+    "衛": 592,
+    "衝": 979,
+    "衡": 1129,
+    "衣": 1317,
+    "表": 77,
+    "衫": 2207,
+    "衰": 2166,
+    "袋": 1477,
+    "袖": 2246,
+    "被": 105,
+    "裁": 1456,
+    "裂": 1067,
+    "裏": 1615,
+    "裔": 1821,
+    "裕": 1666,
+    "裙": 1835,
+    "補": 745,
+    "裝": 469,
+    "裡": 387,
+    "裸": 1601,
+    "裹": 2100,
+    "製": 629,
+    "複": 866,
+    "襪": 1963,
+    "襲": 1883,
+    "西": 94,
+    "要": 40,
+    "覆": 1059,
+    "見": 415,
+    "規": 467,
+    "視": 465,
+    "親": 748,
+    "覲": 1825,
+    "覺": 961,
+    "覽": 1099,
+    "觀": 319,
+    "角": 755,
+    "解": 297,
+    "觸": 957,
+    "言": 299,
+    "訂": 761,
+    "計": 245,
+    "訊": 491,
+    "討": 1430,
+    "訓": 855,
+    "託": 1881,
+    "記": 340,
+    "訝": 2298,
+    "訟": 1486,
+    "訣": 1820,
+    "訪": 973,
+    "設": 471,
+    "許": 123,
+    "訴": 718,
+    "診": 1057,
+    "註": 2243,
+    "詐": 1667,
+    "評": 775,
+    "詛": 2141,
+    "詞": 939,
+    "詢": 1612,
+    "試": 342,
+    "詩": 1658,
+    "話": 505,
+    "該": 93,
+    "詳": 2261,
+    "詹": 1699,
+    "誇": 2232,
+    "誌": 1553,
+    "認": 247,
+    "誕": 1603,
+    "誘": 1828,
+    "語": 152,
+    "誠": 1714,
+    "誤": 1463,
+    "說": 137,
+    "誰": 2130,
+    "課": 1169,
+    "調": 428,
+    "談": 1165,
+    "請": 510,
+    "論": 359,
+    "諮": 2043,
+    "諳": 2077,
+    "諸": 1811,
+    "諺": 1896,
+    "諾": 731,
+    "謀": 1747,
+    "謂": 1903,
+    "謊": 1703,
+    "講": 1696,
+    "謝": 1310,
+    "謹": 1437,
+    "證": 441,
+    "識": 682,
+    "警": 451,
+    "譯": 904,
+    "議": 215,
+    "譴": 1862,
+    "護": 673,
+    "譽": 1226,
+    "讀": 1037,
+    "變": 195,
+    "讓": 274,
+    "讚": 2209,
+    "谷": 1153,
+    "豁": 2124,
+    "豐": 1350,
+    "豔": 2102,
+    "象": 705,
+    "豪": 1336,
+    "豬": 2265,
+    "豹": 927,
+    "貌": 1815,
+    "貓": 1022,
+    "貝": 1229,
+    "負": 801,
+    "財": 1102,
+    "貢": 1634,
+    "貨": 966,
+    "販": 1490,
+    "貪": 1712,
+    "貫": 1579,
+    "責": 1044,
+    "貴": 991,
+    "貶": 2059,
+    "買": 786,
+    "貸": 1483,
+    "費": 414,
+    "貼": 1503,
+    "貿": 1246,
+    "賀": 2134,
+    "資": 382,
+    "賈": 2264,
+    "賊": 2250,
+    "賓": 1166,
+    "賞": 1076,
+    "賠": 2044,
+    "賣": 1208,
+    "質": 353,
+    "賭": 1447,
+    "賴": 1130,
+    "賺": 2002,
+    "購": 663,
+    "賽": 171,
+    "贈": 2144,
+    "贏": 942,
+    "赤": 2198,
+    "赫": 1409,
+    "走": 655,
+    "起": 198,
+    "趁": 2025,
+    "超": 479,
+    "越": 434,
+    "趟": 1640,
+    "趣": 978,
+    "趨": 1920,
+    "足": 390,
+    "跋": 2267,
+    "跌": 1857,
+    "跑": 863,
+    "距": 796,
+    "跟": 595,
+    "跡": 1260,
+    "跨": 908,
+    "路": 233,
+    "跳": 1261,
+    "踐": 1812,
+    "踢": 1305,
+    "蹄": 2064,
+    "蹈": 1851,
+    "蹤": 1663,
+    "躋": 1631,
+    "躍": 1551,
+    "身": 225,
+    "躲": 2291,
+    "車": 109,
+    "軋": 2057,
+    "軌": 1205,
+    "軍": 242,
+    "軟": 1247,
+    "較": 286,
+    "載": 1096,
+    "輔": 1953,
+    "輕": 472,
+    "輛": 1146,
+    "輩": 2293,
+    "輪": 635,
+    "輯": 1241,
+    "輸": 816,
+    "輻": 1532,
+    "轄": 1935,
+    "轉": 648,
+    "轎": 1969,
+    "辛": 1224,
+    "辦": 586,
+    "辨": 1999,
+    "辭": 1944,
+    "辯": 1642,
+    "農": 765,
+    "迄": 2228,
+    "迅": 1346,
+    "迎": 1445,
+    "近": 292,
+    "返": 751,
+    "迦": 2200,
+    "迪": 1031,
+    "迫": 955,
+    "述": 984,
+    "迴": 1084,
+    "迷": 2162,
+    "追": 1161,
+    "退": 1003,
+    "送": 666,
+    "逃": 1526,
+    "逆": 2229,
+    "透": 661,
+    "逐": 1155,
+    "途": 1105,
+    "逗": 1902,
+    "這": 20,
+    "通": 111,
+    "逛": 2183,
+    "逝": 1995,
+    "速": 261,
+    "造": 174,
+    "連": 422,
+    "逮": 1326,
+    "週": 410,
+    "進": 132,
+    "遂": 1757,
+    "遇": 1014,
+    "遊": 281,
+    "運": 189,
+    "遍": 1056,
+    "過": 59,
+    "遏": 2131,
+    "道": 178,
+    "達": 145,
+    "違": 947,
+    "遙": 2085,
+    "遜": 1743,
+    "遠": 522,
+    "遣": 1418,
+    "適": 497,
+    "遭": 487,
+    "遮": 2029,
+    "遲": 1417,
+    "遵": 2020,
+    "遷": 1029,
+    "選": 213,
+    "遺": 714,
+    "避": 691,
+    "邀": 1738,
+    "邁": 1856,
+    "還": 368,
+    "邊": 377,
+    "邏": 2006,
+    "那": 228,
+    "邦": 716,
+    "邸": 1932,
+    "郁": 1869,
+    "郊": 1897,
+    "郡": 2247,
+    "部": 58,
+    "郵": 914,
+    "都": 60,
+    "鄉": 1183,
+    "鄧": 2297,
+    "鄰": 1008,
+    "酋": 2097,
+    "配": 1254,
+    "酒": 548,
+    "酬": 1854,
+    "酸": 1529,
+    "醇": 2116,
+    "醉": 1982,
+    "醒": 813,
+    "醜": 1595,
+    "醫": 504,
+    "釀": 1433,
+    "采": 1722,
+    "釋": 923,
+    "里": 116,
+    "重": 107,
+    "野": 454,
+    "量": 217,
+    "釐": 1455,
+    "金": 294,
+    "針": 968,
+    "鈔": 1802,
+    "鈾": 2084,
+    "鉗": 2121,
+    "鉛": 1754,
+    "銀": 615,
+    "銅": 2208,
+    "銘": 1934,
+    "銜": 1602,
+    "銳": 1744,
+    "銷": 1653,
+    "鋃": 2279,
+    "鋪": 1814,
+    "鋼": 1398,
+    "錄": 711,
+    "錢": 854,
+    "錦": 1509,
+    "錫": 1669,
+    "錯": 1210,
+    "鍊": 1810,
+    "鍋": 1804,
+    "鍛": 1809,
+    "鍵": 1607,
+    "鎊": 1855,
+    "鎖": 1625,
+    "鎢": 2244,
+    "鎮": 1178,
+    "鏡": 453,
+    "鏢": 1250,
+    "鐘": 669,
+    "鐙": 1627,
+    "鐵": 1359,
+    "鐺": 2280,
+    "鑰": 2045,
+    "鑽": 2290,
+    "長": 165,
+    "門": 535,
+    "閃": 1552,
+    "閉": 895,
+    "開": 110,
+    "閒": 1546,
+    "間": 84,
+    "閥": 1530,
+    "閱": 1727,
+    "闊": 1496,
+    "關": 186,
+    "阪": 2214,
+    "防": 656,
+    "阻": 1158,
+    "阿": 316,
+    "陀": 1990,
+    "附": 638,
+    "陌": 2007,
+    "降": 582,
+    "限": 659,
+    "陡": 1075,
+    "院": 283,
+    "陣": 2227,
+    "除": 675,
+    "陪": 1654,
+    "陰": 1838,
+    "陵": 995,
+    "陶": 1805,
+    "陷": 1491,
+    "陸": 587,
+    "陽": 622,
+    "隆": 887,
+    "隊": 167,
+    "階": 758,
+    "隔": 1945,
+    "隕": 2179,
+    "隙": 1479,
+    "際": 369,
+    "障": 1512,
+    "隧": 1657,
+    "隨": 458,
+    "險": 541,
+    "隱": 1116,
+    "隻": 760,
+    "雄": 1614,
+    "雅": 1145,
+    "集": 798,
+    "雇": 1771,
+    "雕": 2038,
+    "雖": 520,
+    "雙": 975,
+    "雜": 892,
+    "雞": 2050,
+    "離": 490,
+    "難": 529,
+    "雨": 781,
+    "雪": 330,
+    "雲": 971,
+    "零": 2226,
+    "雷": 551,
+    "電": 119,
+    "需": 227,
+    "霄": 1311,
+    "震": 879,
+    "霍": 1016,
+    "露": 685,
+    "靈": 1104,
+    "青": 1345,
+    "靜": 1231,
+    "非": 182,
+    "靠": 1174,
+    "面": 172,
+    "革": 1239,
+    "靴": 1841,
+    "鞋": 1319,
+    "鞍": 1733,
+    "鞭": 2098,
+    "韋": 1585,
+    "韓": 1276,
+    "音": 527,
+    "響": 418,
+    "頁": 1292,
+    "頂": 937,
+    "頃": 2170,
+    "項": 538,
+    "順": 1024,
+    "須": 320,
+    "預": 437,
+    "頒": 1281,
+    "頓": 835,
+    "頗": 2051,
+    "領": 459,
+    "頭": 341,
+    "頻": 1225,
+    "顆": 1184,
+    "題": 354,
+    "額": 800,
+    "顎": 2150,
+    "顏": 1977,
+    "願": 1052,
+    "顛": 1334,
+    "類": 173,
+    "顧": 1253,
+    "顯": 500,
+    "顱": 2137,
+    "風": 224,
+    "颱": 1237,
+    "颶": 764,
+    "飛": 380,
+    "食": 446,
+    "飪": 1773,
+    "飯": 778,
+    "飲": 749,
+    "飼": 1639,
+    "飽": 1378,
+    "飾": 1120,
+    "養": 547,
+    "餐": 702,
+    "餘": 1235,
+    "餚": 2013,
+    "館": 777,
+    "餵": 1772,
+    "首": 475,
+    "香": 1269,
+    "馬": 180,
+    "馮": 2145,
+    "馳": 1356,
+    "馴": 1314,
+    "駐": 1150,
+    "駕": 646,
+    "駛": 611,
+    "駝": 1570,
+    "騎": 877,
+    "騙": 1554,
+    "驅": 2014,
+    "驗": 642,
+    "驚": 1196,
+    "骨": 1040,
+    "骸": 2113,
+    "體": 104,
+    "高": 231,
+    "髮": 1872,
+    "鬆": 874,
+    "鬥": 1377,
+    "鬧": 1023,
+    "鬱": 1413,
+    "魁": 2275,
+    "魅": 1605,
+    "魏": 1846,
+    "魚": 996,
+    "魯": 766,
+    "鮮": 1122,
+    "鯊": 1569,
+    "鳥": 1138,
+    "鳴": 1913,
+    "鴉": 1597,
+    "鴨": 2268,
+    "鵝": 1651,
+    "鵡": 2106,
+    "鶇": 1989,
+    "鷹": 1789,
+    "鸚": 2105,
+    "鹿": 938,
+    "麋": 1571,
+    "麗": 1698,
+    "麥": 951,
+    "麻": 1313,
+    "麼": 599,
+    "黃": 1494,
+    "黎": 1560,
+    "黏": 1687,
+    "黑": 665,
+    "默": 1955,
+    "點": 214,
+    "黨": 1164,
+    "鼓": 1923,
+    "鼠": 1706,
+    "鼻": 1680,
+    "齊": 1384,
+    "齋": 1822,
+    "齒": 1510,
+    "齡": 1598,
+    "齧": 2220,
+    "龍": 536,
+    "龐": 1502,
+    "﹑": 1279,
+    "！": 1134,
+    "．": 882,
+    "／": 2180,
+    "：": 253,
+    "；": 579,
+    "？": 784
+  },
+  "yuz": {
+    "'": 45,
+    "-": 24,
+    "0": 37,
+    "1": 38,
+    "2": 39,
+    "3": 42,
+    "4": 41,
+    "5": 46,
+    "6": 43,
+    "7": 44,
+    "8": 49,
+    "9": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 18,
+    "c": 16,
+    "d": 21,
+    "e": 17,
+    "f": 30,
+    "g": 28,
+    "h": 9,
+    "i": 6,
+    "j": 12,
+    "k": 48,
+    "l": 13,
+    "m": 14,
+    "n": 15,
+    "o": 19,
+    "p": 20,
+    "q": 36,
+    "r": 23,
+    "s": 10,
+    "t": 7,
+    "u": 8,
+    "v": 34,
+    "x": 40,
+    "y": 22,
+    "z": 35,
+    "|": 5,
+    "á": 31,
+    "ä": 51,
+    "é": 33,
+    "ë": 11,
+    "í": 29,
+    "ñ": 26,
+    "ó": 32,
+    "ú": 27,
+    "ü": 25,
+    "‐": 47
+  },
+  "yva": {
+    "'": 26,
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 4,
+    "b": 20,
+    "c": 30,
+    "d": 22,
+    "e": 8,
+    "f": 29,
+    "g": 23,
+    "h": 25,
+    "i": 9,
+    "j": 21,
+    "k": 19,
+    "l": 24,
+    "m": 10,
+    "n": 6,
+    "o": 7,
+    "p": 18,
+    "r": 11,
+    "s": 14,
+    "t": 13,
+    "u": 12,
+    "v": 17,
+    "w": 16,
+    "y": 15,
+    "z": 27,
+    "|": 5
+  },
+  "zaa": {
+    "'": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 13,
+    "d": 18,
+    "e": 10,
+    "f": 38,
+    "g": 24,
+    "h": 23,
+    "i": 7,
+    "j": 36,
+    "k": 46,
+    "l": 12,
+    "m": 28,
+    "n": 6,
+    "o": 17,
+    "p": 20,
+    "q": 19,
+    "r": 14,
+    "s": 15,
+    "t": 11,
+    "u": 8,
+    "v": 40,
+    "x": 37,
+    "y": 26,
+    "z": 42,
+    "|": 4,
+    "à": 27,
+    "á": 21,
+    "è": 31,
+    "é": 29,
+    "ì": 30,
+    "í": 25,
+    "ñ": 43,
+    "ò": 44,
+    "ó": 39,
+    "ù": 35,
+    "ú": 34,
+    "ü": 47,
+    "ą": 22,
+    "į": 33,
+    "ı": 49,
+    "̀": 45,
+    "́": 41,
+    "̨": 48,
+    "ẗ": 50,
+    "ỹ": 32
+  },
+  "zab": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 15,
+    "d": 10,
+    "e": 8,
+    "f": 35,
+    "g": 17,
+    "h": 26,
+    "i": 6,
+    "j": 30,
+    "k": 38,
+    "l": 13,
+    "m": 27,
+    "n": 7,
+    "o": 22,
+    "p": 23,
+    "q": 24,
+    "r": 14,
+    "s": 18,
+    "t": 11,
+    "u": 9,
+    "v": 36,
+    "x": 21,
+    "y": 16,
+    "z": 19,
+    "|": 4,
+    "á": 25,
+    "é": 29,
+    "í": 31,
+    "ñ": 39,
+    "ó": 34,
+    "ú": 28,
+    "ü": 32,
+    "ɨ": 20,
+    "́": 37,
+    "—": 33
+  },
+  "zac": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 19,
+    "c": 20,
+    "d": 13,
+    "e": 17,
+    "f": 33,
+    "g": 15,
+    "h": 18,
+    "i": 8,
+    "j": 30,
+    "k": 37,
+    "l": 10,
+    "m": 22,
+    "n": 9,
+    "o": 25,
+    "p": 21,
+    "q": 24,
+    "r": 12,
+    "s": 14,
+    "t": 16,
+    "u": 7,
+    "v": 31,
+    "y": 34,
+    "z": 23,
+    "|": 4,
+    "á": 26,
+    "é": 32,
+    "ë": 5,
+    "í": 28,
+    "ñ": 36,
+    "ó": 35,
+    "ú": 27,
+    "ʼ": 11,
+    "́": 29
+  },
+  "zad": {
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 15,
+    "c": 14,
+    "d": 17,
+    "e": 7,
+    "f": 39,
+    "g": 16,
+    "h": 12,
+    "i": 9,
+    "j": 26,
+    "k": 13,
+    "l": 10,
+    "m": 30,
+    "n": 8,
+    "o": 11,
+    "p": 29,
+    "q": 41,
+    "r": 28,
+    "s": 19,
+    "t": 23,
+    "u": 32,
+    "v": 40,
+    "w": 21,
+    "x": 27,
+    "y": 20,
+    "z": 18,
+    "|": 4,
+    "á": 36,
+    "é": 37,
+    "í": 31,
+    "ñ": 42,
+    "ó": 38,
+    "ú": 34,
+    "ə": 33,
+    "̱": 25,
+    "ḻ": 24,
+    "ṉ": 22,
+    "ꞌ": 6
+  },
+  "zae": {
+    "'": 20,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 25,
+    "d": 18,
+    "e": 7,
+    "f": 33,
+    "g": 17,
+    "h": 14,
+    "i": 8,
+    "j": 31,
+    "k": 13,
+    "l": 11,
+    "m": 29,
+    "n": 6,
+    "o": 19,
+    "p": 30,
+    "q": 37,
+    "r": 15,
+    "s": 16,
+    "t": 9,
+    "u": 10,
+    "v": 35,
+    "w": 27,
+    "x": 21,
+    "y": 22,
+    "z": 36,
+    "|": 4,
+    "á": 23,
+    "é": 32,
+    "í": 24,
+    "ñ": 38,
+    "ó": 34,
+    "ú": 28,
+    "‐": 26
+  },
+  "zai": {
+    "'": 19,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 10,
+    "d": 11,
+    "e": 9,
+    "f": 35,
+    "g": 18,
+    "h": 22,
+    "i": 6,
+    "j": 31,
+    "k": 38,
+    "l": 14,
+    "m": 28,
+    "n": 8,
+    "o": 20,
+    "p": 21,
+    "q": 23,
+    "r": 12,
+    "s": 16,
+    "t": 15,
+    "u": 7,
+    "v": 36,
+    "x": 17,
+    "y": 27,
+    "z": 24,
+    "|": 4,
+    "á": 25,
+    "é": 26,
+    "í": 29,
+    "ñ": 30,
+    "ó": 33,
+    "ú": 32,
+    "ü": 37,
+    "—": 34
+  },
+  "zam": {
+    "'": 27,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 9,
+    "b": 20,
+    "c": 11,
+    "d": 13,
+    "e": 7,
+    "f": 38,
+    "g": 19,
+    "h": 23,
+    "i": 12,
+    "j": 31,
+    "k": 43,
+    "l": 10,
+    "m": 16,
+    "n": 5,
+    "o": 6,
+    "p": 25,
+    "q": 28,
+    "r": 21,
+    "s": 17,
+    "t": 15,
+    "u": 14,
+    "v": 41,
+    "x": 18,
+    "y": 22,
+    "z": 26,
+    "|": 4,
+    "à": 29,
+    "á": 35,
+    "è": 24,
+    "é": 37,
+    "ì": 33,
+    "í": 40,
+    "ñ": 36,
+    "ò": 34,
+    "ó": 30,
+    "ù": 39,
+    "ú": 32,
+    "ü": 42,
+    "ꞌ": 8
+  },
+  "zao": {
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 26,
+    "d": 19,
+    "e": 6,
+    "f": 30,
+    "g": 20,
+    "h": 14,
+    "i": 12,
+    "j": 28,
+    "k": 11,
+    "l": 10,
+    "m": 21,
+    "n": 7,
+    "o": 8,
+    "p": 25,
+    "q": 36,
+    "r": 17,
+    "s": 16,
+    "t": 9,
+    "u": 15,
+    "v": 34,
+    "w": 23,
+    "x": 24,
+    "y": 22,
+    "z": 13,
+    "|": 4,
+    "á": 35,
+    "é": 31,
+    "í": 32,
+    "ó": 33,
+    "ú": 29,
+    "ʼ": 27
+  },
+  "zaq": {
+    "'": 42,
+    "-": 34,
+    "1": 48,
+    "5": 46,
+    "8": 49,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 13,
+    "d": 18,
+    "e": 8,
+    "f": 36,
+    "g": 17,
+    "h": 15,
+    "i": 7,
+    "j": 31,
+    "k": 40,
+    "l": 11,
+    "m": 29,
+    "n": 6,
+    "o": 19,
+    "p": 30,
+    "q": 24,
+    "r": 16,
+    "s": 14,
+    "t": 10,
+    "u": 9,
+    "v": 38,
+    "x": 22,
+    "y": 20,
+    "z": 39,
+    "|": 4,
+    "á": 23,
+    "ä": 32,
+    "é": 35,
+    "í": 25,
+    "ï": 50,
+    "ñ": 44,
+    "ó": 37,
+    "ú": 28,
+    "ü": 33,
+    "̱": 26,
+    "β": 41,
+    "ε": 45,
+    "ι": 47,
+    "ο": 43,
+    "‐": 27,
+    "ꞌ": 21
+  },
+  "zar": {
+    "-": 34,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 14,
+    "c": 12,
+    "d": 16,
+    "e": 13,
+    "f": 37,
+    "g": 15,
+    "h": 28,
+    "i": 7,
+    "j": 29,
+    "l": 10,
+    "m": 36,
+    "n": 9,
+    "o": 25,
+    "p": 33,
+    "q": 26,
+    "r": 19,
+    "s": 21,
+    "t": 17,
+    "u": 8,
+    "v": 39,
+    "x": 23,
+    "y": 27,
+    "z": 20,
+    "|": 4,
+    "á": 22,
+    "é": 31,
+    "ë": 11,
+    "í": 24,
+    "ñ": 41,
+    "ó": 35,
+    "ö": 18,
+    "ú": 30,
+    "ü": 38,
+    "ž": 42,
+    "ǘ": 40,
+    "ʼ": 5,
+    "́": 32
+  },
+  "zas": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 12,
+    "c": 18,
+    "d": 10,
+    "e": 8,
+    "f": 36,
+    "g": 14,
+    "h": 27,
+    "i": 6,
+    "j": 13,
+    "k": 39,
+    "l": 11,
+    "m": 19,
+    "n": 7,
+    "o": 16,
+    "p": 26,
+    "q": 32,
+    "r": 9,
+    "s": 21,
+    "t": 20,
+    "u": 15,
+    "v": 35,
+    "w": 30,
+    "x": 17,
+    "y": 25,
+    "z": 23,
+    "|": 4,
+    "á": 34,
+    "é": 29,
+    "ë": 22,
+    "í": 38,
+    "ñ": 40,
+    "ó": 37,
+    "ú": 33,
+    "ü": 28,
+    "̱": 31
+  },
+  "zav": {
+    "'": 5,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 13,
+    "c": 11,
+    "d": 18,
+    "e": 6,
+    "f": 37,
+    "g": 14,
+    "h": 16,
+    "i": 12,
+    "j": 33,
+    "k": 41,
+    "l": 17,
+    "m": 34,
+    "n": 8,
+    "o": 9,
+    "p": 30,
+    "q": 28,
+    "r": 26,
+    "s": 15,
+    "t": 19,
+    "u": 20,
+    "w": 31,
+    "x": 32,
+    "y": 22,
+    "z": 21,
+    "|": 4,
+    "ñ": 40,
+    "ü": 36,
+    "š": 35,
+    "ž": 25,
+    "ɉ": 29,
+    "ə": 10,
+    "ɟ": 39,
+    "̱": 27,
+    "ḻ": 24,
+    "ṉ": 23,
+    "—": 38
+  },
+  "zaw": {
+    "'": 43,
+    "-": 35,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 20,
+    "c": 12,
+    "d": 13,
+    "e": 8,
+    "f": 34,
+    "g": 18,
+    "h": 23,
+    "i": 5,
+    "j": 10,
+    "k": 44,
+    "l": 14,
+    "m": 29,
+    "n": 7,
+    "o": 19,
+    "p": 26,
+    "q": 32,
+    "r": 11,
+    "s": 15,
+    "t": 17,
+    "u": 9,
+    "v": 42,
+    "w": 31,
+    "x": 16,
+    "y": 28,
+    "z": 24,
+    "|": 4,
+    "á": 41,
+    "ä": 21,
+    "é": 33,
+    "í": 37,
+    "ñ": 45,
+    "ó": 39,
+    "ú": 30,
+    "ü": 46,
+    "́": 40,
+    "̲": 38,
+    "ḻ": 27,
+    "ṉ": 25,
+    "ṟ": 36,
+    "ꞌ": 22
+  },
+  "zca": {
+    "'": 42,
+    "-": 40,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 18,
+    "c": 28,
+    "d": 23,
+    "e": 10,
+    "f": 34,
+    "g": 21,
+    "h": 15,
+    "i": 8,
+    "j": 29,
+    "k": 13,
+    "l": 11,
+    "m": 16,
+    "n": 6,
+    "o": 9,
+    "p": 25,
+    "q": 39,
+    "r": 12,
+    "s": 19,
+    "t": 22,
+    "u": 20,
+    "v": 37,
+    "w": 26,
+    "x": 17,
+    "y": 27,
+    "z": 14,
+    "|": 4,
+    "á": 32,
+    "é": 33,
+    "ë": 24,
+    "í": 36,
+    "ñ": 31,
+    "ó": 35,
+    "ú": 30,
+    "̱": 38,
+    "—": 41,
+    "ꞌ": 7
+  },
+  "zga": {
+    "'": 30,
+    "-": 31,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "d": 23,
+    "e": 8,
+    "f": 29,
+    "g": 16,
+    "h": 26,
+    "i": 9,
+    "j": 25,
+    "k": 12,
+    "l": 10,
+    "m": 15,
+    "n": 7,
+    "o": 14,
+    "p": 22,
+    "s": 19,
+    "t": 18,
+    "u": 20,
+    "v": 11,
+    "w": 17,
+    "y": 21,
+    "z": 28,
+    "|": 4,
+    "ɨ": 13,
+    "ʉ": 6,
+    "ʼ": 27
+  },
+  "zim": {
+    "'": 20,
+    "-": 43,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 24,
+    "d": 18,
+    "e": 13,
+    "f": 23,
+    "g": 16,
+    "h": 21,
+    "i": 6,
+    "j": 27,
+    "k": 10,
+    "l": 17,
+    "m": 8,
+    "n": 7,
+    "o": 11,
+    "p": 28,
+    "r": 14,
+    "s": 15,
+    "t": 12,
+    "u": 9,
+    "v": 25,
+    "w": 22,
+    "y": 30,
+    "z": 29,
+    "|": 4,
+    "à": 26,
+    "á": 40,
+    "â": 31,
+    "é": 42,
+    "ê": 33,
+    "î": 32,
+    "ï": 44,
+    "ô": 34,
+    "ù": 36,
+    "ú": 39,
+    "û": 38,
+    "ʼ": 35,
+    "ˮ": 37,
+    "—": 41
+  },
+  "ziw": {
+    "'": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 23,
+    "c": 22,
+    "d": 19,
+    "e": 8,
+    "f": 26,
+    "g": 17,
+    "h": 16,
+    "i": 6,
+    "j": 25,
+    "k": 12,
+    "l": 14,
+    "m": 11,
+    "n": 7,
+    "o": 13,
+    "p": 27,
+    "s": 18,
+    "t": 20,
+    "u": 9,
+    "v": 24,
+    "w": 10,
+    "y": 15,
+    "z": 21,
+    "|": 4
+  },
+  "zlm": {
+    "!": 67,
+    "$": 53,
+    "%": 51,
+    "&": 55,
+    "'": 47,
+    "+": 65,
+    ",": 41,
+    "-": 27,
+    ".": 37,
+    "/": 43,
+    "0": 28,
+    "1": 30,
+    "2": 32,
+    "3": 38,
+    "4": 35,
+    "5": 34,
+    "6": 39,
+    "7": 40,
+    "8": 36,
+    "9": 33,
+    ":": 46,
+    ";": 52,
+    "<": 48,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    ">": 49,
+    "[": 60,
+    "]": 61,
+    "a": 4,
+    "b": 19,
+    "c": 24,
+    "d": 14,
+    "e": 7,
+    "f": 26,
+    "g": 16,
+    "h": 20,
+    "i": 8,
+    "j": 23,
+    "k": 12,
+    "l": 17,
+    "m": 13,
+    "n": 6,
+    "o": 21,
+    "p": 18,
+    "q": 44,
+    "r": 9,
+    "s": 15,
+    "t": 11,
+    "u": 10,
+    "v": 29,
+    "w": 25,
+    "x": 42,
+    "y": 22,
+    "z": 31,
+    "|": 5,
+    "£": 72,
+    "¥": 54,
+    "°": 59,
+    "²": 68,
+    "á": 50,
+    "ã": 66,
+    "ç": 62,
+    "é": 69,
+    "í": 58,
+    "ó": 70,
+    "õ": 56,
+    "ú": 63,
+    "ü": 57,
+    "ł": 64,
+    "–": 73,
+    "—": 71,
+    "”": 45
+  },
+  "zmz": {
+    "'": 28,
+    "-": 58,
+    "0": 44,
+    "1": 52,
+    "2": 50,
+    "3": 54,
+    "4": 51,
+    "5": 53,
+    "6": 56,
+    "7": 57,
+    "8": 59,
+    "9": 55,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "d": 8,
+    "e": 9,
+    "f": 27,
+    "g": 15,
+    "h": 40,
+    "i": 23,
+    "j": 19,
+    "k": 11,
+    "l": 12,
+    "m": 18,
+    "n": 6,
+    "o": 20,
+    "p": 24,
+    "r": 30,
+    "s": 14,
+    "t": 10,
+    "u": 17,
+    "v": 38,
+    "w": 26,
+    "y": 16,
+    "z": 22,
+    "|": 4,
+    "à": 42,
+    "á": 32,
+    "ã": 39,
+    "è": 46,
+    "é": 41,
+    "ì": 48,
+    "í": 35,
+    "ò": 47,
+    "ó": 37,
+    "õ": 33,
+    "ù": 49,
+    "ú": 36,
+    "ĩ": 34,
+    "œ": 7,
+    "ũ": 60,
+    "ɔ": 21,
+    "ʉ": 25,
+    "̀": 45,
+    "́": 29,
+    "̃": 43,
+    "ẽ": 31
+  },
+  "zne": {
+    "-": 32,
+    "0": 37,
+    "2": 38,
+    "4": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 14,
+    "d": 16,
+    "e": 13,
+    "f": 23,
+    "g": 10,
+    "h": 21,
+    "i": 6,
+    "k": 11,
+    "m": 15,
+    "n": 7,
+    "o": 8,
+    "p": 20,
+    "r": 12,
+    "s": 19,
+    "t": 17,
+    "u": 9,
+    "v": 31,
+    "w": 22,
+    "y": 18,
+    "z": 25,
+    "|": 4,
+    "â": 24,
+    "ã": 28,
+    "ä": 33,
+    "ì": 36,
+    "ò": 27,
+    "ô": 35,
+    "õ": 29,
+    "û": 40,
+    "ĩ": 30,
+    "ũ": 34,
+    "ẽ": 26
+  },
+  "zos": {
+    "'": 9,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 14,
+    "d": 21,
+    "e": 17,
+    "f": 33,
+    "g": 27,
+    "h": 28,
+    "i": 7,
+    "j": 11,
+    "k": 38,
+    "l": 30,
+    "m": 8,
+    "n": 18,
+    "o": 15,
+    "p": 19,
+    "q": 26,
+    "r": 25,
+    "s": 16,
+    "t": 12,
+    "u": 13,
+    "v": 22,
+    "x": 37,
+    "y": 10,
+    "z": 23,
+    "|": 4,
+    "á": 31,
+    "é": 34,
+    "í": 35,
+    "ñ": 29,
+    "ó": 36,
+    "ö": 40,
+    "ø": 6,
+    "ú": 32,
+    "ü": 39,
+    "ṉ": 20
+  },
+  "zpc": {
+    "'": 24,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 13,
+    "c": 17,
+    "d": 15,
+    "e": 7,
+    "f": 35,
+    "g": 21,
+    "h": 26,
+    "i": 6,
+    "j": 22,
+    "k": 40,
+    "l": 12,
+    "m": 29,
+    "n": 10,
+    "o": 14,
+    "p": 27,
+    "q": 25,
+    "r": 16,
+    "s": 23,
+    "t": 19,
+    "u": 9,
+    "v": 38,
+    "x": 28,
+    "y": 20,
+    "z": 18,
+    "|": 4,
+    "­": 41,
+    "á": 32,
+    "é": 33,
+    "ë": 11,
+    "í": 37,
+    "ñ": 34,
+    "ó": 36,
+    "ú": 31,
+    "ü": 39,
+    "̃": 30,
+    "ꞌ": 8
+  },
+  "zpg": {
+    "-": 28,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 22,
+    "c": 23,
+    "d": 10,
+    "e": 5,
+    "f": 42,
+    "g": 17,
+    "h": 18,
+    "i": 6,
+    "j": 36,
+    "k": 48,
+    "l": 12,
+    "m": 15,
+    "n": 8,
+    "o": 9,
+    "p": 26,
+    "q": 31,
+    "r": 16,
+    "s": 25,
+    "t": 20,
+    "u": 13,
+    "v": 44,
+    "w": 29,
+    "x": 24,
+    "y": 11,
+    "z": 14,
+    "|": 4,
+    "á": 43,
+    "â": 27,
+    "é": 34,
+    "ê": 45,
+    "í": 40,
+    "î": 32,
+    "ñ": 19,
+    "ó": 33,
+    "ô": 46,
+    "ú": 38,
+    "û": 47,
+    "ü": 30,
+    "ḻ": 35,
+    "ṉ": 39,
+    "ṟ": 41,
+    "‐": 37,
+    "ꞌ": 21
+  },
+  "zpi": {
+    "-": 30,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 23,
+    "c": 27,
+    "d": 10,
+    "e": 5,
+    "f": 31,
+    "g": 17,
+    "h": 25,
+    "i": 13,
+    "j": 29,
+    "k": 19,
+    "l": 9,
+    "m": 15,
+    "n": 6,
+    "o": 8,
+    "p": 24,
+    "q": 37,
+    "r": 11,
+    "s": 21,
+    "t": 16,
+    "u": 14,
+    "v": 35,
+    "w": 22,
+    "x": 20,
+    "y": 12,
+    "z": 18,
+    "|": 4,
+    "á": 32,
+    "é": 28,
+    "ë": 26,
+    "í": 33,
+    "ñ": 39,
+    "ó": 34,
+    "ú": 38,
+    "́": 36
+  },
+  "zpl": {
+    "'": 11,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 13,
+    "c": 12,
+    "d": 21,
+    "e": 5,
+    "f": 33,
+    "g": 30,
+    "h": 15,
+    "i": 9,
+    "j": 27,
+    "k": 37,
+    "l": 10,
+    "m": 28,
+    "n": 8,
+    "o": 19,
+    "p": 25,
+    "q": 20,
+    "r": 14,
+    "s": 16,
+    "t": 17,
+    "u": 6,
+    "v": 35,
+    "x": 24,
+    "y": 23,
+    "z": 22,
+    "|": 4,
+    "á": 29,
+    "é": 32,
+    "í": 31,
+    "ñ": 26,
+    "ó": 34,
+    "ú": 18,
+    "ü": 36
+  },
+  "zpm": {
+    "'": 42,
+    "-": 39,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 19,
+    "c": 16,
+    "d": 13,
+    "e": 6,
+    "f": 36,
+    "g": 11,
+    "h": 17,
+    "i": 7,
+    "j": 30,
+    "k": 41,
+    "l": 12,
+    "m": 22,
+    "n": 9,
+    "o": 8,
+    "p": 24,
+    "q": 27,
+    "r": 14,
+    "s": 18,
+    "t": 20,
+    "u": 15,
+    "v": 35,
+    "w": 29,
+    "x": 25,
+    "y": 28,
+    "z": 10,
+    "|": 4,
+    "á": 37,
+    "é": 32,
+    "ë": 23,
+    "í": 34,
+    "ñ": 40,
+    "ó": 38,
+    "ú": 31,
+    "ü": 33,
+    "ḻ": 26,
+    "ṉ": 21,
+    "‐": 43
+  },
+  "zpo": {
+    "'": 48,
+    "-": 45,
+    "0": 46,
+    "1": 34,
+    "2": 35,
+    "3": 39,
+    "4": 42,
+    "5": 40,
+    "6": 41,
+    "7": 47,
+    "8": 43,
+    "9": 44,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 16,
+    "c": 25,
+    "d": 23,
+    "e": 6,
+    "f": 31,
+    "g": 24,
+    "h": 15,
+    "i": 12,
+    "j": 27,
+    "k": 13,
+    "l": 8,
+    "m": 22,
+    "n": 7,
+    "o": 9,
+    "p": 26,
+    "q": 37,
+    "r": 17,
+    "s": 20,
+    "t": 19,
+    "u": 11,
+    "v": 36,
+    "w": 21,
+    "x": 10,
+    "y": 18,
+    "z": 14,
+    "|": 4,
+    "á": 38,
+    "é": 28,
+    "í": 33,
+    "ñ": 49,
+    "ó": 32,
+    "ú": 29,
+    "—": 30
+  },
+  "zpt": {
+    "'": 8,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 9,
+    "c": 23,
+    "d": 13,
+    "e": 7,
+    "f": 36,
+    "g": 17,
+    "h": 19,
+    "i": 11,
+    "j": 29,
+    "k": 20,
+    "l": 12,
+    "m": 22,
+    "n": 6,
+    "o": 10,
+    "p": 24,
+    "q": 38,
+    "r": 25,
+    "s": 15,
+    "t": 14,
+    "u": 16,
+    "v": 37,
+    "w": 21,
+    "x": 18,
+    "y": 27,
+    "z": 26,
+    "|": 4,
+    "á": 28,
+    "é": 34,
+    "í": 30,
+    "ñ": 39,
+    "ó": 35,
+    "ú": 31,
+    "‐": 32,
+    "—": 33,
+    "ꞌ": 40
+  },
+  "zpu": {
+    "'": 7,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 8,
+    "b": 13,
+    "c": 21,
+    "d": 15,
+    "e": 5,
+    "f": 36,
+    "g": 25,
+    "h": 19,
+    "i": 12,
+    "j": 22,
+    "k": 11,
+    "l": 9,
+    "m": 29,
+    "n": 6,
+    "o": 10,
+    "p": 26,
+    "q": 38,
+    "r": 14,
+    "s": 17,
+    "t": 18,
+    "u": 28,
+    "v": 37,
+    "w": 23,
+    "x": 24,
+    "y": 16,
+    "z": 20,
+    "|": 4,
+    "á": 30,
+    "é": 27,
+    "í": 34,
+    "ñ": 39,
+    "ó": 35,
+    "ú": 32,
+    "—": 31
+  },
+  "zpz": {
+    "'": 19,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 16,
+    "c": 12,
+    "d": 13,
+    "e": 8,
+    "f": 29,
+    "g": 18,
+    "h": 24,
+    "i": 9,
+    "j": 28,
+    "k": 39,
+    "l": 17,
+    "m": 20,
+    "n": 7,
+    "o": 14,
+    "p": 26,
+    "q": 33,
+    "r": 11,
+    "s": 21,
+    "t": 22,
+    "u": 5,
+    "v": 36,
+    "w": 27,
+    "x": 38,
+    "y": 10,
+    "z": 15,
+    "|": 4,
+    "á": 34,
+    "é": 35,
+    "í": 31,
+    "ñ": 23,
+    "ó": 32,
+    "ú": 37,
+    "ü": 30,
+    "̱": 25
+  },
+  "ztq": {
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 21,
+    "c": 22,
+    "d": 18,
+    "e": 9,
+    "f": 32,
+    "g": 26,
+    "h": 25,
+    "i": 6,
+    "j": 29,
+    "k": 12,
+    "l": 10,
+    "m": 24,
+    "n": 8,
+    "o": 7,
+    "p": 19,
+    "q": 35,
+    "r": 23,
+    "s": 11,
+    "t": 14,
+    "u": 20,
+    "v": 34,
+    "w": 27,
+    "x": 16,
+    "y": 13,
+    "z": 28,
+    "|": 4,
+    "á": 38,
+    "é": 33,
+    "ë": 15,
+    "í": 37,
+    "ñ": 17,
+    "ó": 36,
+    "ú": 30,
+    "—": 31
+  },
+  "zty": {
+    "'": 6,
+    "-": 38,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 11,
+    "c": 18,
+    "d": 16,
+    "e": 5,
+    "f": 40,
+    "g": 13,
+    "h": 19,
+    "i": 12,
+    "j": 25,
+    "k": 9,
+    "l": 17,
+    "m": 37,
+    "n": 8,
+    "o": 10,
+    "p": 34,
+    "r": 33,
+    "s": 22,
+    "t": 20,
+    "u": 35,
+    "w": 21,
+    "x": 30,
+    "y": 23,
+    "z": 24,
+    "|": 4,
+    "á": 26,
+    "é": 28,
+    "í": 32,
+    "ñ": 41,
+    "ó": 29,
+    "ú": 39,
+    "ž": 14,
+    "̱": 31,
+    "ḻ": 15,
+    "ẕ": 27,
+    "—": 36
+  },
+  "zul": {
+    "!": 68,
+    "\"": 53,
+    "$": 47,
+    "%": 49,
+    "&": 58,
+    "'": 45,
+    "+": 61,
+    ",": 43,
+    "-": 24,
+    ".": 41,
+    "/": 46,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "3": 38,
+    "4": 37,
+    "5": 36,
+    "6": 40,
+    "7": 42,
+    "8": 39,
+    "9": 35,
+    ":": 44,
+    ";": 50,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "[": 63,
+    "]": 64,
+    "a": 4,
+    "b": 17,
+    "c": 25,
+    "d": 22,
+    "e": 7,
+    "f": 27,
+    "g": 18,
+    "h": 13,
+    "i": 6,
+    "j": 30,
+    "k": 10,
+    "l": 11,
+    "m": 15,
+    "n": 8,
+    "o": 12,
+    "p": 23,
+    "q": 28,
+    "r": 26,
+    "s": 14,
+    "t": 20,
+    "u": 9,
+    "v": 29,
+    "w": 19,
+    "x": 32,
+    "y": 21,
+    "z": 16,
+    "|": 5,
+    "£": 67,
+    "°": 66,
+    "²": 72,
+    "á": 48,
+    "ã": 65,
+    "ç": 56,
+    "é": 70,
+    "í": 57,
+    "ó": 69,
+    "õ": 59,
+    "ö": 74,
+    "ú": 62,
+    "ü": 51,
+    "ł": 73,
+    "̇": 71,
+    "–": 54,
+    "—": 52,
+    "‘": 60,
+    "”": 55
+  },
+  "zyb": {
+    "'": 30,
+    "-": 33,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 7,
+    "b": 20,
+    "c": 19,
+    "d": 14,
+    "e": 6,
+    "f": 27,
+    "g": 8,
+    "h": 10,
+    "i": 9,
+    "j": 13,
+    "k": 28,
+    "l": 24,
+    "m": 18,
+    "n": 5,
+    "o": 12,
+    "p": 31,
+    "q": 16,
+    "r": 26,
+    "s": 21,
+    "t": 29,
+    "u": 11,
+    "v": 25,
+    "w": 17,
+    "x": 23,
+    "y": 22,
+    "z": 15,
+    "|": 4,
+    "—": 32
+  },
+  "zyp": {
+    "'": 43,
+    "-": 30,
+    "0": 31,
+    "1": 33,
+    "2": 34,
+    "3": 40,
+    "4": 35,
+    "5": 37,
+    "6": 38,
+    "7": 41,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 5,
+    "b": 24,
+    "c": 13,
+    "d": 26,
+    "e": 10,
+    "f": 29,
+    "g": 11,
+    "h": 6,
+    "i": 8,
+    "j": 28,
+    "k": 15,
+    "l": 14,
+    "m": 20,
+    "n": 7,
+    "o": 18,
+    "p": 17,
+    "r": 21,
+    "s": 19,
+    "t": 9,
+    "u": 16,
+    "v": 25,
+    "w": 12,
+    "y": 22,
+    "z": 23,
+    "|": 4,
+    "á": 27,
+    "é": 32,
+    "í": 36,
+    "ó": 42,
+    "ú": 39
+  },
+  "zza": {
+    "'": 36,
+    "-": 37,
+    "</s>": 2,
+    "<pad>": 0,
+    "<s>": 1,
+    "<unk>": 3,
+    "a": 6,
+    "b": 17,
+    "c": 30,
+    "d": 13,
+    "e": 5,
+    "f": 35,
+    "g": 29,
+    "h": 22,
+    "i": 9,
+    "j": 33,
+    "k": 16,
+    "l": 23,
+    "m": 14,
+    "n": 7,
+    "o": 15,
+    "p": 32,
+    "q": 31,
+    "r": 8,
+    "s": 19,
+    "t": 20,
+    "u": 27,
+    "v": 25,
+    "w": 18,
+    "x": 26,
+    "y": 12,
+    "z": 24,
+    "|": 4,
+    "ç": 34,
+    "ê": 11,
+    "î": 10,
+    "û": 28,
+    "ş": 21
+  }
+}