slplab's picture
add tokenizer
b8b9436
raw
history blame
No virus
500 Bytes
{
"A": 0,
"B": 1,
"BB": 2,
"CHh": 3,
"D": 4,
"DD": 5,
"E": 6,
"EO": 7,
"EU": 8,
"G": 9,
"GG": 10,
"H": 11,
"I": 12,
"J": 13,
"JJ": 14,
"Kh": 15,
"L": 16,
"M": 17,
"N": 18,
"NG": 19,
"O": 20,
"Ph": 21,
"R": 22,
"S": 23,
"SS": 24,
"Th": 25,
"U": 26,
"[PAD]": 42,
"[UNK]": 41,
"euI": 27,
"iA": 28,
"iE": 29,
"iEO": 30,
"iO": 31,
"iU": 32,
"k": 33,
"oA": 34,
"oE": 35,
"p": 36,
"t": 37,
"uEO": 38,
"uI": 39,
"|": 40
}