Upload tokenizer
c5e0b41
verified
|
{ |
|
"[PAD]": 101, |
|
"[UNK]": 100, |
|
"a": 1, |
|
"b": 2, |
|
"ba": 3, |
|
"be": 4, |
|
"bi": 5, |
|
"bo": 6, |
|
"bu": 7, |
|
"d": 8, |
|
"da": 9, |
|
"de": 10, |
|
"di": 11, |
|
"do": 12, |
|
"du": 13, |
|
"e": 14, |
|
"f": 15, |
|
"g": 16, |
|
"ga": 17, |
|
"ge": 18, |
|
"gi": 19, |
|
"go": 20, |
|
"gu": 21, |
|
"h": 22, |
|
"ha": 23, |
|
"he": 24, |
|
"hi": 25, |
|
"ho": 26, |
|
"hu": 27, |
|
"i": 28, |
|
"k": 29, |
|
"ka": 30, |
|
"ke": 31, |
|
"ki": 32, |
|
"ko": 33, |
|
"ku": 34, |
|
"l": 35, |
|
"la": 36, |
|
"le": 37, |
|
"li": 38, |
|
"lo": 39, |
|
"lu": 40, |
|
"m": 41, |
|
"n": 42, |
|
"o": 43, |
|
"p": 44, |
|
"pa": 45, |
|
"pe": 46, |
|
"pi": 47, |
|
"po": 48, |
|
"pu": 49, |
|
"r": 50, |
|
"ra": 51, |
|
"re": 52, |
|
"ri": 53, |
|
"ro": 54, |
|
"ru": 55, |
|
"s": 56, |
|
"sa": 57, |
|
"se": 58, |
|
"sh": 59, |
|
"sha": 60, |
|
"she": 61, |
|
"shi": 62, |
|
"sho": 63, |
|
"shu": 64, |
|
"si": 65, |
|
"so": 66, |
|
"su": 67, |
|
"t": 68, |
|
"ta": 69, |
|
"te": 70, |
|
"ti": 71, |
|
"to": 72, |
|
"tu": 73, |
|
"tz": 74, |
|
"u": 75, |
|
"v": 76, |
|
"va": 77, |
|
"ve": 78, |
|
"vi": 79, |
|
"vo": 80, |
|
"vu": 81, |
|
"x": 82, |
|
"xa": 83, |
|
"xe": 84, |
|
"xi": 85, |
|
"xo": 86, |
|
"xu": 87, |
|
"y": 88, |
|
"ya": 89, |
|
"ye": 90, |
|
"yi": 91, |
|
"yo": 92, |
|
"yu": 93, |
|
"z": 94, |
|
"za": 95, |
|
"ze": 96, |
|
"zi": 97, |
|
"zo": 98, |
|
"zu": 99, |
|
"|": 0 |
|
} |
|
|