Mengkedalai's picture
Upload tokenizer
c048d0a verified
raw
history blame
488 Bytes
{
"'": 28,
"[PAD]": 40,
"[UNK]": 39,
"h": 18,
"|": 9,
"«": 4,
"»": 6,
"а": 34,
"б": 35,
"в": 12,
"г": 30,
"д": 27,
"е": 36,
"ж": 24,
"з": 15,
"и": 1,
"й": 20,
"к": 17,
"л": 3,
"м": 11,
"н": 2,
"о": 13,
"п": 14,
"р": 21,
"с": 22,
"т": 29,
"у": 0,
"ф": 38,
"х": 16,
"ц": 19,
"ч": 8,
"ш": 37,
"ъ": 5,
"ы": 33,
"ь": 25,
"э": 7,
"ю": 31,
"я": 26,
"ё": 10,
"ү": 32,
"ө": 23
}