timotewos's picture
Upload tokenizer
abad9d3 verified
raw
history blame
3.12 kB
{
"[PAD]": 229,
"[UNK]": 228,
"|": 187,
"แˆ€": 141,
"แˆ": 12,
"แˆ‚": 113,
"แˆƒ": 219,
"แˆ„": 181,
"แˆ…": 221,
"แˆ†": 178,
"แˆˆ": 78,
"แˆ‰": 4,
"แˆŠ": 16,
"แˆ‹": 86,
"แˆŒ": 139,
"แˆ": 15,
"แˆŽ": 24,
"แˆ": 81,
"แˆ": 76,
"แˆ‘": 64,
"แˆ“": 44,
"แˆ”": 171,
"แˆ•": 45,
"แˆ˜": 74,
"แˆ™": 124,
"แˆš": 191,
"แˆ›": 199,
"แˆœ": 94,
"แˆ": 183,
"แˆž": 84,
"แˆŸ": 157,
"แˆ ": 35,
"แˆก": 72,
"แˆฃ": 222,
"แˆฅ": 140,
"แˆฆ": 132,
"แˆจ": 151,
"แˆฉ": 225,
"แˆช": 226,
"แˆซ": 133,
"แˆฌ": 122,
"แˆญ": 147,
"แˆฎ": 189,
"แˆฏ": 188,
"แˆฐ": 214,
"แˆฑ": 150,
"แˆฒ": 68,
"แˆณ": 42,
"แˆด": 63,
"แˆต": 53,
"แˆถ": 27,
"แˆท": 208,
"แˆธ": 128,
"แˆน": 21,
"แˆบ": 43,
"แˆป": 160,
"แˆผ": 186,
"แˆฝ": 201,
"แˆพ": 227,
"แ‰€": 129,
"แ‰": 200,
"แ‰‚": 9,
"แ‰ƒ": 179,
"แ‰„": 105,
"แ‰…": 115,
"แ‰†": 101,
"แ‰‹": 206,
"แ‰ ": 111,
"แ‰ก": 50,
"แ‰ข": 60,
"แ‰ฃ": 10,
"แ‰ค": 223,
"แ‰ฅ": 217,
"แ‰ฆ": 185,
"แ‰ง": 167,
"แ‰จ": 103,
"แ‰ช": 184,
"แ‰ซ": 158,
"แ‰ญ": 195,
"แ‰ฎ": 107,
"แ‰ฐ": 47,
"แ‰ฑ": 177,
"แ‰ฒ": 163,
"แ‰ณ": 51,
"แ‰ด": 146,
"แ‰ต": 137,
"แ‰ถ": 159,
"แ‰ท": 98,
"แ‰ธ": 148,
"แ‰น": 93,
"แ‰บ": 212,
"แ‰ป": 2,
"แ‰ผ": 62,
"แ‰ฝ": 117,
"แ‰พ": 92,
"แ‰ฟ": 20,
"แŠ€": 210,
"แŠƒ": 156,
"แŠ…": 73,
"แŠ‹": 7,
"แŠ": 56,
"แŠ‘": 134,
"แŠ’": 190,
"แŠ“": 57,
"แŠ”": 65,
"แŠ•": 164,
"แŠ–": 216,
"แŠ—": 55,
"แŠ˜": 215,
"แŠ™": 22,
"แŠ›": 207,
"แŠ": 83,
"แŠž": 144,
"แŠŸ": 61,
"แŠ ": 1,
"แŠก": 3,
"แŠข": 202,
"แŠค": 192,
"แŠฅ": 48,
"แŠฆ": 58,
"แŠจ": 99,
"แŠฉ": 87,
"แŠช": 135,
"แŠซ": 66,
"แŠฌ": 32,
"แŠญ": 96,
"แŠฎ": 168,
"แŠณ": 29,
"แŠธ": 224,
"แŠฝ": 174,
"แ‹ˆ": 198,
"แ‹Š": 69,
"แ‹‹": 108,
"แ‹Œ": 19,
"แ‹": 49,
"แ‹Ž": 138,
"แ‹": 209,
"แ‹‘": 127,
"แ‹’": 95,
"แ‹“": 166,
"แ‹•": 97,
"แ‹–": 118,
"แ‹˜": 172,
"แ‹™": 33,
"แ‹š": 70,
"แ‹›": 0,
"แ‹œ": 13,
"แ‹": 161,
"แ‹ž": 38,
"แ‹Ÿ": 119,
"แ‹ข": 46,
"แ‹ฃ": 52,
"แ‹ค": 114,
"แ‹ฅ": 169,
"แ‹ฆ": 30,
"แ‹จ": 125,
"แ‹ฉ": 220,
"แ‹ซ": 25,
"แ‹ฌ": 89,
"แ‹ญ": 6,
"แ‹ฎ": 175,
"แ‹ฐ": 106,
"แ‹ฑ": 162,
"แ‹ฒ": 91,
"แ‹ณ": 123,
"แ‹ด": 152,
"แ‹ต": 102,
"แ‹ถ": 88,
"แ‹ท": 204,
"แŒ€": 218,
"แŒ": 149,
"แŒ‚": 71,
"แŒƒ": 126,
"แŒ„": 11,
"แŒ…": 41,
"แŒ†": 40,
"แŒˆ": 176,
"แŒ‰": 180,
"แŒŠ": 143,
"แŒ‹": 14,
"แŒŒ": 131,
"แŒ": 120,
"แŒŽ": 110,
"แŒ“": 213,
"แŒ ": 194,
"แŒก": 8,
"แŒข": 136,
"แŒฃ": 142,
"แŒค": 153,
"แŒฅ": 121,
"แŒฆ": 112,
"แŒง": 5,
"แŒจ": 155,
"แŒฉ": 116,
"แŒช": 205,
"แŒซ": 85,
"แŒฌ": 79,
"แŒญ": 154,
"แŒฎ": 67,
"แŒฒ": 193,
"แŒด": 31,
"แŒต": 145,
"แŒถ": 18,
"แŒธ": 90,
"แŒน": 36,
"แŒบ": 173,
"แŒป": 104,
"แŒฝ": 23,
"แŒพ": 77,
"แŒฟ": 203,
"แ€": 54,
"แ": 37,
"แƒ": 109,
"แ…": 80,
"แˆ": 28,
"แ‰": 34,
"แŠ": 82,
"แ‹": 196,
"แŒ": 17,
"แ": 59,
"แŽ": 39,
"แ": 26,
"แ‘": 165,
"แ’": 130,
"แ“": 197,
"แ”": 75,
"แ•": 211,
"แ–": 100,
"โ€œ": 170,
"โ€": 182
}