auditi41's picture
Upload tokenizer
ecaa06b
raw
history blame
1.02 kB
{
"'": 39,
"/": 21,
"[PAD]": 79,
"[UNK]": 78,
"a": 20,
"e": 9,
"g": 72,
"l": 63,
"o": 67,
"p": 34,
"y": 53,
"|": 15,
"।": 55,
"॥": 18,
"ঁ": 26,
"ং": 44,
"ঃ": 69,
"অ": 7,
"আ": 33,
"ই": 38,
"ঈ": 8,
"উ": 30,
"ঊ": 65,
"ঋ": 46,
"এ": 60,
"ঐ": 19,
"ও": 47,
"ঔ": 42,
"ক": 14,
"খ": 31,
"গ": 29,
"ঘ": 24,
"ঙ": 48,
"চ": 13,
"ছ": 1,
"জ": 51,
"ঝ": 52,
"ঞ": 4,
"ট": 71,
"ঠ": 64,
"ড": 58,
"ঢ": 35,
"ণ": 66,
"ত": 28,
"থ": 37,
"দ": 70,
"ধ": 49,
"ন": 54,
"প": 62,
"ফ": 59,
"ব": 3,
"ভ": 12,
"ম": 41,
"য": 50,
"র": 25,
"ল": 32,
"শ": 40,
"ষ": 45,
"স": 27,
"হ": 22,
"়": 10,
"া": 57,
"ি": 43,
"ী": 0,
"ু": 5,
"ূ": 17,
"ৃ": 74,
"ে": 16,
"ৈ": 76,
"ো": 68,
"ৌ": 36,
"্": 73,
"ৎ": 11,
"ড়": 23,
"ঢ়": 2,
"য়": 56,
"ৰ": 6,
"—": 77,
"’": 61,
"‚": 75
}