AdaCodruta's picture
Upload tokenizer
96e6e1c verified
raw
history blame
547 Bytes
{
" ": 25,
"[PAD]": 43,
"[UNK]": 42,
"a": 21,
"b": 37,
"c": 15,
"d": 12,
"e": 32,
"f": 9,
"g": 28,
"h": 7,
"i": 29,
"j": 38,
"k": 40,
"l": 20,
"m": 8,
"n": 0,
"o": 2,
"p": 10,
"q": 17,
"r": 31,
"s": 26,
"t": 23,
"u": 33,
"v": 22,
"w": 39,
"x": 16,
"y": 41,
"z": 27,
"á": 18,
"â": 3,
"é": 34,
"î": 11,
"ü": 4,
"ă": 5,
"ć": 35,
"č": 24,
"đ": 36,
"ş": 19,
"š": 1,
"ţ": 6,
"ș": 14,
"ț": 13,
"”": 30
}