tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
97c7c0c
raw
history blame
814 Bytes
{
"!": 24,
"\"": 60,
"'": 29,
"(": 13,
")": 52,
",": 30,
"-": 25,
".": 59,
":": 43,
";": 54,
"?": 7,
"A": 19,
"B": 20,
"C": 28,
"D": 17,
"E": 53,
"F": 14,
"G": 32,
"H": 49,
"I": 66,
"J": 11,
"K": 16,
"L": 44,
"M": 40,
"N": 6,
"O": 48,
"P": 51,
"R": 58,
"S": 12,
"T": 1,
"U": 47,
"V": 38,
"Z": 67,
"[PAD]": 71,
"[UNK]": 70,
"a": 42,
"b": 37,
"c": 18,
"d": 55,
"e": 65,
"f": 31,
"g": 35,
"h": 50,
"i": 9,
"j": 69,
"k": 27,
"l": 56,
"m": 2,
"n": 46,
"o": 5,
"p": 41,
"q": 3,
"r": 26,
"s": 10,
"t": 15,
"u": 23,
"v": 39,
"w": 57,
"y": 45,
"z": 8,
"|": 21,
"Β«": 33,
"Δ΄": 62,
"–": 22,
"β€”": 63,
"β€˜": 4,
"’": 34,
"β€œ": 61,
"”": 64,
"β€ž": 68,
"fi": 0,
"fl": 36
}