xlsr-nm-clp / vocab.json
susmitabhatt's picture
Upload tokenizer
1434fb9 verified
raw
history blame
309 Bytes
{
"[PAD]": 27,
"[UNK]": 26,
"a": 1,
"b": 7,
"c": 4,
"d": 12,
"e": 3,
"f": 22,
"g": 8,
"h": 0,
"i": 16,
"j": 14,
"k": 13,
"l": 9,
"m": 17,
"n": 20,
"o": 2,
"p": 21,
"r": 25,
"s": 15,
"t": 6,
"u": 10,
"v": 11,
"w": 18,
"x": 24,
"y": 23,
"z": 5,
"|": 19
}