cdactvm's picture
Upload tokenizer
80dd778 verified
raw
history blame contribute delete
853 Bytes
{
"[PAD]": 65,
"[UNK]": 64,
"|": 0,
"।": 1,
"ঁ": 2,
"ং": 3,
"ঃ": 4,
"অ": 5,
"আ": 6,
"ই": 7,
"ঈ": 8,
"উ": 9,
"এ": 10,
"ঐ": 11,
"ও": 12,
"ঔ": 13,
"ক": 14,
"খ": 15,
"গ": 16,
"ঘ": 17,
"ঙ": 18,
"চ": 19,
"ছ": 20,
"জ": 21,
"ঝ": 22,
"ঞ": 23,
"ট": 24,
"ঠ": 25,
"ড": 26,
"ঢ": 27,
"ণ": 28,
"ত": 29,
"থ": 30,
"দ": 31,
"ধ": 32,
"ন": 33,
"প": 34,
"ফ": 35,
"ব": 36,
"ভ": 37,
"ম": 38,
"য": 39,
"র": 40,
"ল": 41,
"শ": 42,
"ষ": 43,
"স": 44,
"হ": 45,
"়": 46,
"া": 47,
"ি": 48,
"ী": 49,
"ু": 50,
"ূ": 51,
"ৃ": 52,
"ে": 53,
"ৈ": 54,
"ো": 55,
"ৌ": 56,
"্": 57,
"ৎ": 58,
"ড়": 59,
"ঢ়": 60,
"য়": 61,
"ৰ": 62,
"ৱ": 63
}