wav2vec2-ru / vocab.json
Alikhan Urumov
add tokenizer
f4403d2
raw
history blame
430 Bytes
{
"[PAD]": 35,
"[UNK]": 34,
"|": 20,
"а": 12,
"б": 14,
"в": 24,
"г": 22,
"д": 23,
"е": 10,
"ж": 15,
"з": 30,
"и": 1,
"й": 8,
"к": 27,
"л": 9,
"м": 11,
"н": 31,
"о": 21,
"п": 5,
"р": 3,
"с": 7,
"т": 0,
"у": 2,
"ф": 6,
"х": 17,
"ц": 28,
"ч": 33,
"ш": 26,
"щ": 13,
"ъ": 16,
"ы": 32,
"ь": 4,
"э": 19,
"ю": 29,
"я": 18,
"ё": 25
}