XLSR-LM-NewData / vocab.json
gokul-g-menon's picture
Upload tokenizer
368feec verified
raw
history blame contribute delete
983 Bytes
{
"[PAD]": 75,
"[UNK]": 74,
"|": 0,
"ം": 1,
"ഃ": 2,
"അ": 3,
"ആ": 4,
"ഇ": 5,
"ഈ": 6,
"ഉ": 7,
"ഊ": 8,
"ഋ": 9,
"എ": 10,
"ഏ": 11,
"ഐ": 12,
"ഒ": 13,
"ഓ": 14,
"ഔ": 15,
"ക": 16,
"ഖ": 17,
"ഗ": 18,
"ഘ": 19,
"ങ": 20,
"ച": 21,
"ഛ": 22,
"ജ": 23,
"ഝ": 24,
"ഞ": 25,
"ട": 26,
"ഠ": 27,
"ഡ": 28,
"ഢ": 29,
"ണ": 30,
"ത": 31,
"ഥ": 32,
"ദ": 33,
"ധ": 34,
"ന": 35,
"പ": 36,
"ഫ": 37,
"ബ": 38,
"ഭ": 39,
"മ": 40,
"യ": 41,
"ര": 42,
"റ": 43,
"ല": 44,
"ള": 45,
"ഴ": 46,
"വ": 47,
"ശ": 48,
"ഷ": 49,
"സ": 50,
"ഹ": 51,
"ാ": 52,
"ി": 53,
"ീ": 54,
"ു": 55,
"ൂ": 56,
"ൃ": 57,
"െ": 58,
"േ": 59,
"ൈ": 60,
"ൊ": 61,
"ോ": 62,
"ൌ": 63,
"്": 64,
"ൗ": 65,
"ൺ": 66,
"ൻ": 67,
"ർ": 68,
"ൽ": 69,
"ൾ": 70,
"‌": 71,
"‍": 72,
"‎": 73
}