ssc-qxp-mms-model-bl / vocab.json
ctaguchi's picture
Upload tokenizer
34e9e53 verified
{
"qxp": {
"!": 7,
"'": 22,
",": 35,
"-": 18,
".": 51,
"...": 69,
";": 68,
"?": 12,
"A": 26,
"B": 10,
"C": 14,
"Ch": 11,
"D": 67,
"E": 48,
"G": 33,
"H": 43,
"I": 36,
"J": 52,
"K": 29,
"L": 20,
"Ll": 25,
"M": 66,
"N": 6,
"P": 49,
"Q": 16,
"R": 23,
"S": 9,
"Sh": 24,
"T": 60,
"U": 32,
"V": 8,
"W": 31,
"Y": 47,
"[PAD]": 71,
"[UNK]": 70,
"a": 38,
"b": 17,
"c": 42,
"ch": 1,
"d": 59,
"e": 2,
"f": 27,
"g": 56,
"h": 55,
"i": 44,
"j": 65,
"k": 53,
"l": 58,
"ll": 54,
"m": 64,
"n": 19,
"o": 5,
"p": 39,
"q": 41,
"r": 46,
"s": 50,
"sh": 61,
"t": 28,
"u": 4,
"v": 57,
"w": 13,
"y": 34,
"z": 30,
"|": 15,
"¡": 45,
"¿": 0,
"Ñ": 37,
"á": 63,
"í": 21,
"ñ": 62,
"ú": 40,
"’": 3
}
}