krishivoice / vocab.json
krishi0311's picture
Upload tokenizer
c0c90bf verified
raw
history blame
320 Bytes
{
"[PAD]": 28,
"[UNK]": 27,
"a": 21,
"b": 12,
"c": 8,
"d": 4,
"e": 25,
"f": 16,
"g": 15,
"h": 9,
"i": 3,
"j": 23,
"k": 26,
"l": 14,
"m": 1,
"n": 18,
"o": 5,
"p": 17,
"q": 20,
"r": 11,
"s": 10,
"t": 6,
"u": 7,
"v": 0,
"w": 2,
"x": 22,
"y": 13,
"z": 24,
"|": 19
}