speech31's picture
add tokenizer
1a61a0c
raw
history blame
499 Bytes
{
"'": 16,
"[PAD]": 43,
"[UNK]": 42,
"a": 40,
"b": 30,
"c": 4,
"d": 2,
"e": 1,
"f": 27,
"g": 23,
"h": 26,
"i": 28,
"j": 0,
"k": 32,
"l": 3,
"m": 9,
"n": 24,
"o": 38,
"p": 29,
"q": 34,
"r": 11,
"s": 15,
"t": 35,
"u": 6,
"v": 41,
"w": 22,
"x": 37,
"y": 12,
"z": 13,
"|": 8,
"æ": 33,
"ð": 14,
"ŋ": 7,
"ɑ": 31,
"ɔ": 25,
"ə": 17,
"ɛ": 5,
"ɪ": 10,
"ʃ": 18,
"ʊ": 39,
"ʒ": 21,
"ʤ": 20,
"ʧ": 36,
"θ": 19
}