colerobertson's picture
Upload tokenizer
9003acd verified
raw
history blame
564 Bytes
{
"[PAD]": 0,
"[UNK]": 48,
"a": 26,
"b": 9,
"d": 39,
"e": 34,
"f": 38,
"h": 16,
"i": 24,
"j": 10,
"k": 7,
"l": 25,
"m": 8,
"n": 12,
"o": 32,
"p": 2,
"r": 28,
"s": 22,
"t": 6,
"u": 23,
"w": 47,
"x": 35,
"y": 33,
"z": 17,
"|": 44,
"ç": 18,
"ŋ": 20,
"ɑ": 19,
"ɒ": 40,
"ɔ": 14,
"ə": 29,
"ɛ": 46,
"ɜ": 27,
"ɡ": 30,
"ɪ": 15,
"ɸ": 45,
"ɹ": 21,
"ɻ": 42,
"ʂ": 37,
"ʃ": 1,
"ʊ": 4,
"ʋ": 43,
"ʌ": 11,
"ʒ": 36,
"ʔ": 5,
"ʘ": 41,
"ʝ": 3,
"β": 13,
"θ": 31
}