ad019el's picture
Upload tokenizer
611b3e2
raw
history blame
554 Bytes
{
"[PAD]": 47,
"[UNK]": 46,
"a": 31,
"b": 8,
"c": 27,
"d": 30,
"e": 17,
"f": 37,
"g": 26,
"h": 4,
"i": 43,
"j": 24,
"k": 45,
"l": 36,
"m": 5,
"n": 21,
"o": 39,
"p": 6,
"q": 10,
"r": 16,
"s": 9,
"t": 7,
"u": 15,
"v": 18,
"w": 12,
"x": 44,
"y": 11,
"z": 0,
"|": 19,
" ": 35,
"ĉ": 34,
"č": 3,
"ğ": 32,
"ţ": 22,
"ǧ": 25,
"ɛ": 2,
"ɣ": 33,
"̣": 14,
"γ": 20,
"ε": 40,
"ԑ": 1,
"چ": 42,
"ḍ": 13,
"ḥ": 28,
"ṛ": 29,
"ṣ": 23,
"ṭ": 38,
"ẓ": 41
}