sanchit-gandhi's picture
Upload tokenizer
87191bc
raw
history blame
397 Bytes
{
" ": 11,
"'": 17,
"-": 23,
"_": 10,
"a": 1,
"b": 27,
"c": 2,
"d": 15,
"e": 8,
"f": 20,
"g": 28,
"h": 26,
"i": 24,
"j": 22,
"k": 19,
"l": 12,
"m": 0,
"n": 14,
"o": 13,
"p": 16,
"q": 3,
"r": 5,
"s": 29,
"t": 34,
"u": 31,
"v": 32,
"w": 21,
"x": 7,
"y": 6,
"z": 33,
"ā": 18,
"ē": 9,
"ī": 30,
"ō": 25,
"‐": 35,
"—": 4
}