mms-tts-mon / vocab.json
sanchit-gandhi's picture
Upload tokenizer
e015410
raw
history blame contribute delete
733 Bytes
{
" ": 63,
"-": 62,
"a": 59,
"b": 39,
"c": 51,
"d": 52,
"e": 38,
"f": 43,
"g": 46,
"h": 61,
"i": 41,
"j": 36,
"k": 54,
"l": 37,
"m": 42,
"n": 44,
"o": 56,
"p": 40,
"q": 48,
"r": 47,
"s": 50,
"t": 60,
"u": 55,
"v": 49,
"w": 53,
"x": 45,
"y": 57,
"z": 58,
"|": 0,
"а": 1,
"б": 16,
"в": 20,
"г": 4,
"д": 6,
"е": 25,
"ж": 21,
"з": 23,
"и": 7,
"й": 12,
"к": 33,
"л": 10,
"м": 18,
"н": 3,
"о": 13,
"п": 32,
"р": 5,
"с": 15,
"т": 11,
"у": 14,
"ф": 34,
"х": 8,
"ц": 27,
"ч": 19,
"щ": 26,
"ъ": 35,
"ы": 24,
"ь": 22,
"э": 2,
"ю": 29,
"я": 28,
"ё": 31,
"ѳ": 17,
"ү": 9,
"—": 30
}