w2v-bert-2.0-odia_v1 / vocab.json
cdactvm's picture
Upload tokenizer
479e40b verified
raw
history blame contribute delete
892 Bytes
{
"[PAD]": 68,
"[UNK]": 67,
"|": 0,
"।": 1,
"ଁ": 2,
"ଂ": 3,
"ଃ": 4,
"ଅ": 5,
"ଆ": 6,
"ଇ": 7,
"ଈ": 8,
"ଉ": 9,
"ଊ": 10,
"ଋ": 11,
"ଏ": 12,
"ଐ": 13,
"ଓ": 14,
"ଔ": 15,
"କ": 16,
"ଖ": 17,
"ଗ": 18,
"ଘ": 19,
"ଙ": 20,
"ଚ": 21,
"ଛ": 22,
"ଜ": 23,
"ଝ": 24,
"ଞ": 25,
"ଟ": 26,
"ଠ": 27,
"ଡ": 28,
"ଢ": 29,
"ଣ": 30,
"ତ": 31,
"ଥ": 32,
"ଦ": 33,
"ଧ": 34,
"ନ": 35,
"ପ": 36,
"ଫ": 37,
"ବ": 38,
"ଭ": 39,
"ମ": 40,
"ଯ": 41,
"ର": 42,
"ଲ": 43,
"ଳ": 44,
"ଵ": 45,
"ଶ": 46,
"ଷ": 47,
"ସ": 48,
"ହ": 49,
"଼": 50,
"ା": 51,
"ି": 52,
"ୀ": 53,
"ୁ": 54,
"ୂ": 55,
"ୃ": 56,
"େ": 57,
"ୈ": 58,
"ୋ": 59,
"ୌ": 60,
"୍": 61,
"ଡ଼": 62,
"ଢ଼": 63,
"ୟ": 64,
"ୱ": 65,
"‌": 66
}