ad019el's picture
Upload tokenizer
fb8c22c
raw
history blame contribute delete
554 Bytes
{
"[PAD]": 47,
"[UNK]": 46,
"a": 1,
"b": 22,
"c": 37,
"d": 21,
"e": 2,
"f": 26,
"g": 34,
"h": 5,
"i": 44,
"j": 45,
"k": 28,
"l": 9,
"m": 18,
"n": 36,
"o": 38,
"p": 10,
"q": 40,
"r": 11,
"s": 33,
"t": 6,
"u": 24,
"v": 41,
"w": 32,
"x": 20,
"y": 7,
"z": 29,
"|": 15,
" ": 30,
"ĉ": 42,
"č": 27,
"ğ": 3,
"ţ": 0,
"ǧ": 14,
"ɛ": 25,
"ɣ": 43,
"̣": 35,
"γ": 4,
"ε": 39,
"ԑ": 23,
"چ": 19,
"ḍ": 8,
"ḥ": 12,
"ṛ": 31,
"ṣ": 16,
"ṭ": 17,
"ẓ": 13
}