audio1 / vocab.json
thanhnew2001's picture
Upload tokenizer
734a492 verified
raw
history blame contribute delete
487 Bytes
{
" ": 42,
"'": 41,
"-": 2,
"`": 37,
"a": 14,
"c": 10,
"e": 9,
"f": 29,
"g": 6,
"h": 3,
"i": 4,
"k": 5,
"l": 16,
"m": 17,
"n": 1,
"o": 12,
"p": 19,
"s": 11,
"t": 7,
"u": 8,
"v": 30,
"y": 15,
"|": 0,
"à": 25,
"á": 28,
"â": 18,
"è": 34,
"é": 35,
"ê": 20,
"ì": 13,
"í": 36,
"î": 24,
"ò": 26,
"ó": 21,
"ô": 31,
"ù": 33,
"ú": 32,
"û": 23,
"̀": 39,
"́": 38,
"̂": 40,
"̍": 27,
"ṳ": 22
}