xlsr-aiish-clp / vocab.json
susmitabhatt's picture
Upload tokenizer
d6d0259 verified
raw
history blame
309 Bytes
{
"D": 17,
"K": 3,
"O": 15,
"T": 5,
"[PAD]": 27,
"[UNK]": 26,
"a": 25,
"b": 8,
"c": 10,
"d": 2,
"e": 21,
"g": 20,
"h": 24,
"i": 13,
"j": 0,
"k": 12,
"l": 7,
"m": 4,
"n": 18,
"o": 16,
"p": 14,
"r": 23,
"s": 19,
"t": 1,
"u": 11,
"v": 9,
"y": 6,
"|": 22
}