xlsr-aiish-clp / vocab.json
susmitabhatt's picture
Upload tokenizer
62dc5c8 verified
raw
history blame
No virus
309 Bytes
{
"D": 15,
"K": 5,
"O": 18,
"T": 10,
"[PAD]": 27,
"[UNK]": 26,
"a": 20,
"b": 17,
"c": 21,
"d": 11,
"e": 12,
"g": 22,
"h": 7,
"i": 13,
"j": 23,
"k": 0,
"l": 14,
"m": 6,
"n": 3,
"o": 16,
"p": 2,
"r": 19,
"s": 8,
"t": 1,
"u": 9,
"v": 24,
"y": 25,
"|": 4
}