sanchit-gandhi's picture
Upload tokenizer
80e2619
raw
history blame
526 Bytes
{
" ": 44,
"0": 36,
"1": 37,
"2": 38,
"3": 42,
"4": 39,
"5": 43,
"6": 40,
"|": 0,
"а": 1,
"б": 9,
"в": 24,
"г": 8,
"д": 6,
"е": 19,
"ж": 30,
"з": 16,
"и": 2,
"й": 20,
"к": 15,
"л": 4,
"м": 11,
"н": 3,
"о": 7,
"п": 28,
"р": 5,
"с": 13,
"т": 12,
"у": 10,
"ф": 31,
"х": 23,
"ч": 22,
"ш": 18,
"ъ": 35,
"ь": 41,
"э": 25,
"ю": 34,
"я": 27,
"ё": 26,
"ў": 17,
"ғ": 29,
"қ": 14,
"ҳ": 21,
"–": 32,
"—": 33
}