Upload tokenizer
33d9861
verified
|
{ |
|
"mal": { |
|
"[PAD]": 72, |
|
"[UNK]": 71, |
|
"|": 0, |
|
"ം": 1, |
|
"ഃ": 2, |
|
"അ": 3, |
|
"ആ": 4, |
|
"ഇ": 5, |
|
"ഈ": 6, |
|
"ഉ": 7, |
|
"ഊ": 8, |
|
"എ": 9, |
|
"ഏ": 10, |
|
"ഐ": 11, |
|
"ഒ": 12, |
|
"ഓ": 13, |
|
"ഔ": 14, |
|
"ക": 15, |
|
"ഖ": 16, |
|
"ഗ": 17, |
|
"ഘ": 18, |
|
"ങ": 19, |
|
"ച": 20, |
|
"ഛ": 21, |
|
"ജ": 22, |
|
"ഞ": 23, |
|
"ട": 24, |
|
"ഠ": 25, |
|
"ഡ": 26, |
|
"ഢ": 27, |
|
"ണ": 28, |
|
"ത": 29, |
|
"ഥ": 30, |
|
"ദ": 31, |
|
"ധ": 32, |
|
"ന": 33, |
|
"പ": 34, |
|
"ഫ": 35, |
|
"ബ": 36, |
|
"ഭ": 37, |
|
"മ": 38, |
|
"യ": 39, |
|
"ര": 40, |
|
"റ": 41, |
|
"ല": 42, |
|
"ള": 43, |
|
"ഴ": 44, |
|
"വ": 45, |
|
"ശ": 46, |
|
"ഷ": 47, |
|
"സ": 48, |
|
"ഹ": 49, |
|
"ാ": 50, |
|
"ി": 51, |
|
"ീ": 52, |
|
"ു": 53, |
|
"ൂ": 54, |
|
"ൃ": 55, |
|
"െ": 56, |
|
"േ": 57, |
|
"ൈ": 58, |
|
"ൊ": 59, |
|
"ോ": 60, |
|
"ൌ": 61, |
|
"്": 62, |
|
"ൗ": 63, |
|
"ൺ": 64, |
|
"ൻ": 65, |
|
"ർ": 66, |
|
"ൽ": 67, |
|
"ൾ": 68, |
|
"ൿ": 69, |
|
"’": 70 |
|
}, |
|
"tam": { |
|
"&": 1, |
|
"[PAD]": 71, |
|
"[UNK]": 70, |
|
"_": 2, |
|
"`": 3, |
|
"|": 0, |
|
"·": 4, |
|
"½": 5, |
|
"¾": 6, |
|
"á": 7, |
|
"é": 8, |
|
"ô": 9, |
|
"š": 10, |
|
"ஃ": 11, |
|
"அ": 12, |
|
"ஆ": 13, |
|
"இ": 14, |
|
"ஈ": 15, |
|
"உ": 16, |
|
"ஊ": 17, |
|
"எ": 18, |
|
"ஏ": 19, |
|
"ஐ": 20, |
|
"ஒ": 21, |
|
"ஓ": 22, |
|
"ஔ": 23, |
|
"க": 24, |
|
"ங": 25, |
|
"ச": 26, |
|
"ஜ": 27, |
|
"ஞ": 28, |
|
"ட": 29, |
|
"ண": 30, |
|
"த": 31, |
|
"ந": 32, |
|
"ன": 33, |
|
"ப": 34, |
|
"ம": 35, |
|
"ய": 36, |
|
"ர": 37, |
|
"ற": 38, |
|
"ல": 39, |
|
"ள": 40, |
|
"ழ": 41, |
|
"வ": 42, |
|
"ஷ": 43, |
|
"ஸ": 44, |
|
"ஹ": 45, |
|
"ா": 46, |
|
"ி": 47, |
|
"ீ": 48, |
|
"ு": 49, |
|
"ூ": 50, |
|
"ெ": 51, |
|
"ே": 52, |
|
"ை": 53, |
|
"ொ": 54, |
|
"ோ": 55, |
|
"ௌ": 56, |
|
"்": 57, |
|
"ௗ": 58, |
|
"ഥ": 59, |
|
"–": 60, |
|
"—": 61, |
|
"’": 62, |
|
"‚": 63, |
|
"•": 64, |
|
"…": 65, |
|
"′": 66, |
|
"″": 67, |
|
"●": 68, |
|
"◯": 69 |
|
} |
|
} |
|
|