Upload tokenizer
722dec7
verified
|
{ |
|
" ": 84, |
|
"&": 45, |
|
"'": 39, |
|
"(": 46, |
|
")": 83, |
|
"[PAD]": 101, |
|
"[UNK]": 100, |
|
"\\": 88, |
|
"_": 86, |
|
"`": 6, |
|
"a": 42, |
|
"b": 24, |
|
"c": 80, |
|
"d": 37, |
|
"e": 5, |
|
"f": 85, |
|
"g": 29, |
|
"h": 62, |
|
"i": 72, |
|
"j": 70, |
|
"k": 25, |
|
"l": 56, |
|
"m": 10, |
|
"n": 57, |
|
"o": 89, |
|
"p": 31, |
|
"q": 92, |
|
"r": 98, |
|
"s": 95, |
|
"t": 59, |
|
"u": 58, |
|
"v": 74, |
|
"w": 63, |
|
"x": 33, |
|
"y": 16, |
|
"z": 94, |
|
"·": 40, |
|
"½": 81, |
|
"¾": 1, |
|
"á": 71, |
|
"é": 51, |
|
"ô": 90, |
|
"š": 76, |
|
"ஃ": 60, |
|
"அ": 43, |
|
"ஆ": 65, |
|
"இ": 64, |
|
"ஈ": 15, |
|
"உ": 19, |
|
"ஊ": 0, |
|
"எ": 22, |
|
"ஏ": 41, |
|
"ஐ": 3, |
|
"ஒ": 97, |
|
"ஓ": 28, |
|
"ஔ": 2, |
|
"க": 17, |
|
"ங": 23, |
|
"ச": 93, |
|
"ஜ": 8, |
|
"ஞ": 87, |
|
"ட": 75, |
|
"ண": 18, |
|
"த": 14, |
|
"ந": 66, |
|
"ன": 82, |
|
"ப": 96, |
|
"ம": 30, |
|
"ய": 11, |
|
"ர": 38, |
|
"ற": 35, |
|
"ல": 44, |
|
"ள": 73, |
|
"ழ": 55, |
|
"வ": 79, |
|
"ஷ": 52, |
|
"ஸ": 26, |
|
"ஹ": 99, |
|
"ா": 50, |
|
"ி": 32, |
|
"ீ": 54, |
|
"ு": 20, |
|
"ூ": 48, |
|
"ெ": 47, |
|
"ே": 21, |
|
"ை": 4, |
|
"ொ": 7, |
|
"ோ": 49, |
|
"ௌ": 12, |
|
"்": 34, |
|
"ௗ": 27, |
|
"ഥ": 69, |
|
"–": 36, |
|
"—": 9, |
|
"’": 61, |
|
"‚": 68, |
|
"•": 77, |
|
"…": 78, |
|
"′": 67, |
|
"″": 53, |
|
"●": 91, |
|
"◯": 13 |
|
} |
|
|