wav2vec2-Malayalam / vocab.json
Bluecast's picture
Upload tokenizer
ded9b9b verified
raw
history blame contribute delete
955 Bytes
{
"'": 69,
"[PAD]": 73,
"[UNK]": 72,
"|": 31,
"ം": 25,
"ഃ": 6,
"അ": 55,
"ആ": 11,
"ഇ": 27,
"ഈ": 60,
"ഉ": 68,
"ഊ": 43,
"എ": 38,
"ഏ": 71,
"ഐ": 46,
"ഒ": 18,
"ഓ": 47,
"ഔ": 51,
"ക": 64,
"ഖ": 44,
"ഗ": 9,
"ഘ": 52,
"ങ": 61,
"ച": 19,
"ഛ": 7,
"ജ": 32,
"ഞ": 62,
"ട": 29,
"ഠ": 53,
"ഡ": 42,
"ഢ": 23,
"ണ": 67,
"ത": 8,
"ഥ": 24,
"ദ": 16,
"ധ": 48,
"ന": 36,
"പ": 28,
"ഫ": 5,
"ബ": 35,
"ഭ": 4,
"മ": 65,
"യ": 26,
"ര": 30,
"റ": 13,
"ല": 22,
"ള": 17,
"ഴ": 3,
"വ": 21,
"ശ": 10,
"ഷ": 63,
"സ": 54,
"ഹ": 15,
"ാ": 40,
"ി": 0,
"ീ": 34,
"ു": 59,
"ൂ": 57,
"ൃ": 70,
"െ": 1,
"േ": 33,
"ൈ": 41,
"ൊ": 58,
"ോ": 12,
"ൌ": 56,
"്": 14,
"ൗ": 49,
"ൺ": 39,
"ൻ": 45,
"ർ": 50,
"ൽ": 2,
"ൾ": 66,
"ൿ": 37,
"’": 20
}