Upload tokenizer
3e487d2
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]": 53, |
|
"[UNK]": 52, |
|
"_": 2, |
|
"|": 0, |
|
"ஃ": 3, |
|
"அ": 4, |
|
"ஆ": 5, |
|
"இ": 6, |
|
"ஈ": 7, |
|
"உ": 8, |
|
"ஊ": 9, |
|
"எ": 10, |
|
"ஏ": 11, |
|
"ஐ": 12, |
|
"ஒ": 13, |
|
"ஓ": 14, |
|
"ஔ": 15, |
|
"க": 16, |
|
"ங": 17, |
|
"ச": 18, |
|
"ஜ": 19, |
|
"ஞ": 20, |
|
"ட": 21, |
|
"ண": 22, |
|
"த": 23, |
|
"ந": 24, |
|
"ன": 25, |
|
"ப": 26, |
|
"ம": 27, |
|
"ய": 28, |
|
"ர": 29, |
|
"ற": 30, |
|
"ல": 31, |
|
"ள": 32, |
|
"ழ": 33, |
|
"வ": 34, |
|
"ஷ": 35, |
|
"ஸ": 36, |
|
"ஹ": 37, |
|
"ா": 38, |
|
"ி": 39, |
|
"ீ": 40, |
|
"ு": 41, |
|
"ூ": 42, |
|
"ெ": 43, |
|
"ே": 44, |
|
"ை": 45, |
|
"ொ": 46, |
|
"ோ": 47, |
|
"ௌ": 48, |
|
"்": 49, |
|
"ௗ": 50, |
|
"ഥ": 51 |
|
}, |
|
"tel": { |
|
"[PAD]": 53, |
|
"[UNK]": 52, |
|
"|": 0, |
|
"ం": 1, |
|
"అ": 2, |
|
"ఆ": 3, |
|
"ఇ": 4, |
|
"ఈ": 5, |
|
"ఉ": 6, |
|
"ఊ": 7, |
|
"ఎ": 8, |
|
"ఏ": 9, |
|
"ఒ": 10, |
|
"క": 11, |
|
"ఖ": 12, |
|
"గ": 13, |
|
"ఘ": 14, |
|
"చ": 15, |
|
"జ": 16, |
|
"ట": 17, |
|
"డ": 18, |
|
"ణ": 19, |
|
"త": 20, |
|
"థ": 21, |
|
"ద": 22, |
|
"ధ": 23, |
|
"న": 24, |
|
"ప": 25, |
|
"ఫ": 26, |
|
"బ": 27, |
|
"భ": 28, |
|
"మ": 29, |
|
"య": 30, |
|
"ర": 31, |
|
"ల": 32, |
|
"ళ": 33, |
|
"వ": 34, |
|
"శ": 35, |
|
"ష": 36, |
|
"స": 37, |
|
"హ": 38, |
|
"ా": 39, |
|
"ి": 40, |
|
"ీ": 41, |
|
"ు": 42, |
|
"ూ": 43, |
|
"ృ": 44, |
|
"ె": 45, |
|
"ే": 46, |
|
"ై": 47, |
|
"ొ": 48, |
|
"ో": 49, |
|
"ౌ": 50, |
|
"్": 51 |
|
}, |
|
"yor": { |
|
"[PAD]": 43, |
|
"[UNK]": 42, |
|
"a": 1, |
|
"b": 2, |
|
"d": 3, |
|
"e": 4, |
|
"f": 5, |
|
"g": 6, |
|
"h": 7, |
|
"i": 8, |
|
"j": 9, |
|
"k": 10, |
|
"l": 11, |
|
"m": 12, |
|
"n": 13, |
|
"o": 14, |
|
"p": 15, |
|
"r": 16, |
|
"s": 17, |
|
"t": 18, |
|
"u": 19, |
|
"w": 20, |
|
"y": 21, |
|
"|": 0, |
|
"à": 22, |
|
"á": 23, |
|
"è": 24, |
|
"é": 25, |
|
"ì": 26, |
|
"í": 27, |
|
"ò": 28, |
|
"ó": 29, |
|
"ù": 30, |
|
"ú": 31, |
|
"ń": 32, |
|
"ǹ": 33, |
|
"̀": 34, |
|
"́": 35, |
|
"̄": 36, |
|
"̣": 37, |
|
"ṣ": 38, |
|
"ẹ": 39, |
|
"ọ": 40, |
|
"ụ": 41 |
|
} |
|
} |
|
|
