mms-tts-hoc / vocab.json
sanchit-gandhi's picture
Upload tokenizer
c376b69
raw
history blame
No virus
696 Bytes
{
" ": 5,
"'": 7,
"-": 9,
"4": 54,
"_": 39,
"i": 42,
"ଁ": 34,
"ଂ": 26,
"ଃ": 20,
"ଅ": 23,
"ଆ": 44,
"ଇ": 28,
"ଉ": 31,
"ଏ": 51,
"କ": 21,
"ଖ": 27,
"ଗ": 52,
"ଙ": 30,
"ଚ": 45,
"ଜ": 53,
"ଞ": 17,
"ଟ": 0,
"ଡ": 25,
"ଢ": 1,
"ଣ": 29,
"ତ": 15,
"ଥ": 8,
"ଦ": 40,
"ନ": 14,
"ପ": 48,
"ଫ": 4,
"ବ": 47,
"ଭ": 32,
"ମ": 6,
"ର": 38,
"ଲ": 50,
"ଳ": 49,
"ଶ": 24,
"ସ": 2,
"ହ": 22,
"ା": 46,
"ି": 3,
"ୀ": 35,
"ୁ": 43,
"ୃ": 11,
"େ": 18,
"ୈ": 13,
"ୋ": 37,
"ୌ": 12,
"୍": 10,
"ୟ": 16,
"ୱ": 41,
"‍": 33,
"–": 19,
"—": 36
}