w2v2-Malasar / vocab.json
kavyamanohar's picture
Upload tokenizer
cc4fd1b verified
raw
history blame
593 Bytes
{
"[PAD]": 45,
"[UNK]": 44,
"|": 0,
"அ": 1,
"ஆ": 2,
"இ": 3,
"ஈ": 4,
"உ": 5,
"ஊ": 6,
"எ": 7,
"ஏ": 8,
"ஐ": 9,
"ஒ": 10,
"ஓ": 11,
"க": 12,
"ங": 13,
"ச": 14,
"ஜ": 15,
"ஞ": 16,
"ட": 17,
"ண": 18,
"த": 19,
"ந": 20,
"ன": 21,
"ப": 22,
"ம": 23,
"ய": 24,
"ர": 25,
"ற": 26,
"ல": 27,
"ள": 28,
"ழ": 29,
"வ": 30,
"ஷ": 31,
"ஸ": 32,
"ா": 33,
"ி": 34,
"ீ": 35,
"ு": 36,
"ூ": 37,
"ெ": 38,
"ே": 39,
"ை": 40,
"ொ": 41,
"ோ": 42,
"்": 43
}