nbaden's picture
Upload tokenizer
3194be8
raw
history blame
546 Bytes
{
"[PAD]": 45,
"[UNK]": 44,
"c": 34,
"i": 42,
"o": 9,
"y": 16,
"|": 8,
"«": 0,
"»": 12,
"а": 40,
"б": 25,
"в": 39,
"г": 24,
"д": 21,
"е": 29,
"ж": 6,
"з": 33,
"и": 7,
"й": 13,
"к": 10,
"л": 31,
"м": 23,
"н": 22,
"о": 11,
"п": 20,
"р": 15,
"с": 1,
"т": 32,
"у": 5,
"ф": 26,
"х": 35,
"ц": 41,
"ч": 38,
"ш": 4,
"щ": 37,
"ы": 43,
"ь": 28,
"э": 3,
"ю": 17,
"я": 30,
"і": 18,
"ҕ": 27,
"ҥ": 14,
"ү": 36,
"һ": 2,
"ө": 19
}