DipsankarSinha's picture
Upload tokenizer
0541e1d verified
raw
history blame
3.12 kB
{
"[PAD]": 229,
"[UNK]": 228,
"|": 171,
"แˆ€": 127,
"แˆ": 140,
"แˆ‚": 211,
"แˆƒ": 199,
"แˆ„": 44,
"แˆ…": 86,
"แˆ†": 180,
"แˆˆ": 142,
"แˆ‰": 105,
"แˆŠ": 213,
"แˆ‹": 32,
"แˆŒ": 111,
"แˆ": 172,
"แˆŽ": 67,
"แˆ": 207,
"แˆ": 158,
"แˆ‘": 63,
"แˆ“": 159,
"แˆ”": 60,
"แˆ•": 118,
"แˆ˜": 203,
"แˆ™": 54,
"แˆš": 16,
"แˆ›": 57,
"แˆœ": 119,
"แˆ": 121,
"แˆž": 90,
"แˆŸ": 154,
"แˆ ": 4,
"แˆก": 209,
"แˆฃ": 128,
"แˆฅ": 175,
"แˆฆ": 164,
"แˆจ": 191,
"แˆฉ": 101,
"แˆช": 227,
"แˆซ": 122,
"แˆฌ": 12,
"แˆญ": 89,
"แˆฎ": 76,
"แˆฏ": 45,
"แˆฐ": 178,
"แˆฑ": 146,
"แˆฒ": 166,
"แˆณ": 13,
"แˆด": 219,
"แˆต": 80,
"แˆถ": 107,
"แˆท": 109,
"แˆธ": 28,
"แˆน": 169,
"แˆบ": 198,
"แˆป": 167,
"แˆผ": 78,
"แˆฝ": 36,
"แˆพ": 37,
"แ‰€": 5,
"แ‰": 216,
"แ‰‚": 145,
"แ‰ƒ": 70,
"แ‰„": 147,
"แ‰…": 38,
"แ‰†": 94,
"แ‰‹": 185,
"แ‰ ": 88,
"แ‰ก": 71,
"แ‰ข": 168,
"แ‰ฃ": 20,
"แ‰ค": 184,
"แ‰ฅ": 33,
"แ‰ฆ": 17,
"แ‰ง": 6,
"แ‰จ": 218,
"แ‰ช": 222,
"แ‰ซ": 27,
"แ‰ญ": 196,
"แ‰ฎ": 160,
"แ‰ฐ": 64,
"แ‰ฑ": 188,
"แ‰ฒ": 141,
"แ‰ณ": 62,
"แ‰ด": 96,
"แ‰ต": 72,
"แ‰ถ": 52,
"แ‰ท": 77,
"แ‰ธ": 40,
"แ‰น": 42,
"แ‰บ": 197,
"แ‰ป": 176,
"แ‰ผ": 14,
"แ‰ฝ": 92,
"แ‰พ": 34,
"แ‰ฟ": 215,
"แŠ€": 144,
"แŠƒ": 59,
"แŠ…": 148,
"แŠ‹": 217,
"แŠ": 104,
"แŠ‘": 49,
"แŠ’": 156,
"แŠ“": 46,
"แŠ”": 26,
"แŠ•": 25,
"แŠ–": 43,
"แŠ—": 192,
"แŠ˜": 162,
"แŠ™": 100,
"แŠ›": 68,
"แŠ": 206,
"แŠž": 190,
"แŠŸ": 132,
"แŠ ": 61,
"แŠก": 113,
"แŠข": 30,
"แŠค": 47,
"แŠฅ": 2,
"แŠฆ": 125,
"แŠจ": 117,
"แŠฉ": 69,
"แŠช": 31,
"แŠซ": 134,
"แŠฌ": 75,
"แŠญ": 194,
"แŠฎ": 173,
"แŠณ": 83,
"แŠธ": 138,
"แŠฝ": 201,
"แ‹ˆ": 106,
"แ‹Š": 195,
"แ‹‹": 150,
"แ‹Œ": 135,
"แ‹": 10,
"แ‹Ž": 139,
"แ‹": 53,
"แ‹‘": 1,
"แ‹’": 73,
"แ‹“": 200,
"แ‹•": 126,
"แ‹–": 220,
"แ‹˜": 205,
"แ‹™": 11,
"แ‹š": 87,
"แ‹›": 19,
"แ‹œ": 93,
"แ‹": 130,
"แ‹ž": 55,
"แ‹Ÿ": 48,
"แ‹ข": 212,
"แ‹ฃ": 22,
"แ‹ค": 8,
"แ‹ฅ": 221,
"แ‹ฆ": 187,
"แ‹จ": 226,
"แ‹ฉ": 112,
"แ‹ซ": 202,
"แ‹ฌ": 182,
"แ‹ญ": 123,
"แ‹ฎ": 51,
"แ‹ฐ": 223,
"แ‹ฑ": 174,
"แ‹ฒ": 208,
"แ‹ณ": 3,
"แ‹ด": 177,
"แ‹ต": 143,
"แ‹ถ": 189,
"แ‹ท": 137,
"แŒ€": 179,
"แŒ": 18,
"แŒ‚": 84,
"แŒƒ": 163,
"แŒ„": 15,
"แŒ…": 225,
"แŒ†": 115,
"แŒˆ": 35,
"แŒ‰": 81,
"แŒŠ": 65,
"แŒ‹": 157,
"แŒŒ": 74,
"แŒ": 204,
"แŒŽ": 103,
"แŒ“": 124,
"แŒ ": 29,
"แŒก": 110,
"แŒข": 129,
"แŒฃ": 50,
"แŒค": 56,
"แŒฅ": 131,
"แŒฆ": 165,
"แŒง": 23,
"แŒจ": 214,
"แŒฉ": 186,
"แŒช": 24,
"แŒซ": 152,
"แŒฌ": 193,
"แŒญ": 114,
"แŒฎ": 39,
"แŒฒ": 9,
"แŒด": 116,
"แŒต": 210,
"แŒถ": 21,
"แŒธ": 41,
"แŒน": 66,
"แŒบ": 170,
"แŒป": 151,
"แŒฝ": 85,
"แŒพ": 82,
"แŒฟ": 79,
"แ€": 181,
"แ": 133,
"แƒ": 149,
"แ…": 153,
"แˆ": 98,
"แ‰": 7,
"แŠ": 224,
"แ‹": 95,
"แŒ": 136,
"แ": 97,
"แŽ": 183,
"แ": 120,
"แ‘": 99,
"แ’": 155,
"แ“": 161,
"แ”": 108,
"แ•": 0,
"แ–": 91,
"โ€œ": 58,
"โ€": 102
}