add tokenizer
9c2fe18
|
{ |
|
"1": 82, |
|
"2": 94, |
|
"3": 12, |
|
"4": 97, |
|
"5": 81, |
|
"7": 48, |
|
"8": 68, |
|
"A": 77, |
|
"B": 79, |
|
"C": 98, |
|
"D": 100, |
|
"E": 62, |
|
"F": 26, |
|
"G": 16, |
|
"H": 23, |
|
"I": 58, |
|
"J": 91, |
|
"K": 20, |
|
"L": 88, |
|
"M": 102, |
|
"N": 86, |
|
"O": 37, |
|
"P": 42, |
|
"Q": 80, |
|
"R": 15, |
|
"S": 70, |
|
"T": 17, |
|
"U": 11, |
|
"V": 108, |
|
"W": 29, |
|
"X": 107, |
|
"Y": 14, |
|
"Z": 9, |
|
"[PAD]": 111, |
|
"[UNK]": 110, |
|
"a": 78, |
|
"b": 2, |
|
"c": 67, |
|
"d": 59, |
|
"e": 69, |
|
"f": 21, |
|
"g": 53, |
|
"h": 33, |
|
"i": 89, |
|
"j": 47, |
|
"k": 71, |
|
"l": 57, |
|
"m": 13, |
|
"n": 56, |
|
"o": 34, |
|
"p": 61, |
|
"q": 19, |
|
"r": 36, |
|
"s": 27, |
|
"t": 64, |
|
"u": 93, |
|
"v": 39, |
|
"w": 25, |
|
"x": 32, |
|
"y": 38, |
|
"z": 40, |
|
"|": 24, |
|
"γ±": 43, |
|
"γ²": 46, |
|
"γ³": 22, |
|
"γ΄": 103, |
|
"γ΅": 109, |
|
"γΆ": 74, |
|
"γ·": 41, |
|
"γΈ": 76, |
|
"γΉ": 83, |
|
"γΊ": 66, |
|
"γ»": 51, |
|
"γΌ": 4, |
|
"γΎ": 105, |
|
"γΏ": 31, |
|
"γ
": 87, |
|
"γ
": 52, |
|
"γ
": 55, |
|
"γ
": 35, |
|
"γ
": 85, |
|
"γ
": 92, |
|
"γ
": 50, |
|
"γ
": 7, |
|
"γ
": 90, |
|
"γ
": 84, |
|
"γ
": 8, |
|
"γ
": 72, |
|
"γ
": 96, |
|
"γ
": 60, |
|
"γ
": 44, |
|
"γ
": 95, |
|
"γ
": 54, |
|
"γ
": 0, |
|
"γ
": 101, |
|
"γ
": 49, |
|
"γ
": 45, |
|
"γ
": 75, |
|
"γ
": 106, |
|
"γ
": 63, |
|
"γ
": 10, |
|
"γ
": 104, |
|
"γ
": 6, |
|
"γ
": 3, |
|
"γ
": 1, |
|
"γ
": 99, |
|
"γ
": 18, |
|
"γ
": 30, |
|
"γ
": 28, |
|
"γ
‘": 5, |
|
"γ
’": 65, |
|
"γ
£": 73 |
|
} |
|
|