mms-tts-coe / vocab.json
sanchit-gandhi's picture
Upload tokenizer
5ce8f25
raw
history blame contribute delete
471 Bytes
{
" ": 41,
"'": 4,
"a": 0,
"b": 31,
"c": 6,
"d": 22,
"e": 3,
"f": 34,
"g": 32,
"h": 8,
"i": 5,
"j": 11,
"k": 15,
"l": 27,
"m": 13,
"n": 9,
"o": 7,
"p": 14,
"q": 36,
"r": 10,
"s": 12,
"t": 17,
"u": 16,
"v": 20,
"x": 40,
"y": 39,
"z": 38,
"|": 1,
"á": 35,
"ã": 18,
"é": 33,
"í": 30,
"ñ": 19,
"ó": 37,
"õ": 28,
"ú": 29,
"ĩ": 25,
"ũ": 21,
"ʉ": 2,
"̃": 23,
"ẽ": 24,
"—": 26
}