ad019el's picture
Upload tokenizer
a245345
raw
history blame
453 Bytes
{
"[PAD]": 38,
"[UNK]": 37,
"a": 33,
"b": 19,
"c": 20,
"e": 0,
"f": 5,
"g": 23,
"j": 18,
"k": 28,
"l": 34,
"m": 15,
"o": 1,
"q": 26,
"r": 10,
"t": 32,
"u": 29,
"w": 31,
"x": 27,
"y": 25,
"z": 4,
"|": 35,
"ĉ": 13,
"č": 24,
"ğ": 36,
"ţ": 7,
"ǧ": 22,
"ɛ": 17,
"ɣ": 30,
"̣": 11,
"γ": 14,
"ε": 6,
"ԑ": 9,
"ḍ": 12,
"ḥ": 2,
"ṛ": 21,
"ṣ": 8,
"ṭ": 3,
"ẓ": 16
}