timotewos's picture
Upload tokenizer
64075f8 verified
raw
history blame
3.12 kB
{
"[PAD]": 229,
"[UNK]": 228,
"|": 161,
"แˆ€": 219,
"แˆ": 162,
"แˆ‚": 67,
"แˆƒ": 191,
"แˆ„": 125,
"แˆ…": 80,
"แˆ†": 184,
"แˆˆ": 53,
"แˆ‰": 109,
"แˆŠ": 103,
"แˆ‹": 36,
"แˆŒ": 183,
"แˆ": 46,
"แˆŽ": 139,
"แˆ": 17,
"แˆ": 1,
"แˆ‘": 130,
"แˆ“": 227,
"แˆ”": 16,
"แˆ•": 209,
"แˆ˜": 200,
"แˆ™": 11,
"แˆš": 85,
"แˆ›": 152,
"แˆœ": 35,
"แˆ": 135,
"แˆž": 173,
"แˆŸ": 143,
"แˆ ": 165,
"แˆก": 12,
"แˆฃ": 22,
"แˆฅ": 91,
"แˆฆ": 97,
"แˆจ": 30,
"แˆฉ": 19,
"แˆช": 29,
"แˆซ": 79,
"แˆฌ": 0,
"แˆญ": 101,
"แˆฎ": 142,
"แˆฏ": 27,
"แˆฐ": 26,
"แˆฑ": 132,
"แˆฒ": 20,
"แˆณ": 107,
"แˆด": 45,
"แˆต": 178,
"แˆถ": 167,
"แˆท": 43,
"แˆธ": 48,
"แˆน": 86,
"แˆบ": 62,
"แˆป": 24,
"แˆผ": 83,
"แˆฝ": 55,
"แˆพ": 168,
"แ‰€": 116,
"แ‰": 75,
"แ‰‚": 214,
"แ‰ƒ": 192,
"แ‰„": 126,
"แ‰…": 5,
"แ‰†": 215,
"แ‰‹": 28,
"แ‰ ": 188,
"แ‰ก": 98,
"แ‰ข": 13,
"แ‰ฃ": 50,
"แ‰ค": 3,
"แ‰ฅ": 39,
"แ‰ฆ": 151,
"แ‰ง": 124,
"แ‰จ": 34,
"แ‰ช": 181,
"แ‰ซ": 68,
"แ‰ญ": 102,
"แ‰ฎ": 194,
"แ‰ฐ": 96,
"แ‰ฑ": 6,
"แ‰ฒ": 61,
"แ‰ณ": 74,
"แ‰ด": 153,
"แ‰ต": 158,
"แ‰ถ": 222,
"แ‰ท": 185,
"แ‰ธ": 41,
"แ‰น": 207,
"แ‰บ": 15,
"แ‰ป": 52,
"แ‰ผ": 224,
"แ‰ฝ": 145,
"แ‰พ": 51,
"แ‰ฟ": 37,
"แŠ€": 147,
"แŠƒ": 160,
"แŠ…": 197,
"แŠ‹": 58,
"แŠ": 129,
"แŠ‘": 204,
"แŠ’": 44,
"แŠ“": 104,
"แŠ”": 93,
"แŠ•": 180,
"แŠ–": 221,
"แŠ—": 88,
"แŠ˜": 174,
"แŠ™": 100,
"แŠ›": 226,
"แŠ": 176,
"แŠž": 40,
"แŠŸ": 89,
"แŠ ": 141,
"แŠก": 206,
"แŠข": 31,
"แŠค": 70,
"แŠฅ": 179,
"แŠฆ": 66,
"แŠจ": 8,
"แŠฉ": 205,
"แŠช": 32,
"แŠซ": 123,
"แŠฌ": 170,
"แŠญ": 156,
"แŠฎ": 127,
"แŠณ": 150,
"แŠธ": 225,
"แŠฝ": 189,
"แ‹ˆ": 198,
"แ‹Š": 187,
"แ‹‹": 154,
"แ‹Œ": 69,
"แ‹": 164,
"แ‹Ž": 108,
"แ‹": 117,
"แ‹‘": 216,
"แ‹’": 54,
"แ‹“": 56,
"แ‹•": 105,
"แ‹–": 47,
"แ‹˜": 90,
"แ‹™": 202,
"แ‹š": 140,
"แ‹›": 212,
"แ‹œ": 113,
"แ‹": 171,
"แ‹ž": 81,
"แ‹Ÿ": 155,
"แ‹ข": 201,
"แ‹ฃ": 131,
"แ‹ค": 120,
"แ‹ฅ": 23,
"แ‹ฆ": 159,
"แ‹จ": 134,
"แ‹ฉ": 14,
"แ‹ซ": 182,
"แ‹ฌ": 138,
"แ‹ญ": 213,
"แ‹ฎ": 78,
"แ‹ฐ": 10,
"แ‹ฑ": 128,
"แ‹ฒ": 121,
"แ‹ณ": 190,
"แ‹ด": 196,
"แ‹ต": 169,
"แ‹ถ": 49,
"แ‹ท": 59,
"แŒ€": 157,
"แŒ": 149,
"แŒ‚": 203,
"แŒƒ": 136,
"แŒ„": 7,
"แŒ…": 57,
"แŒ†": 114,
"แŒˆ": 137,
"แŒ‰": 146,
"แŒŠ": 65,
"แŒ‹": 217,
"แŒŒ": 223,
"แŒ": 25,
"แŒŽ": 110,
"แŒ“": 64,
"แŒ ": 175,
"แŒก": 193,
"แŒข": 133,
"แŒฃ": 92,
"แŒค": 71,
"แŒฅ": 2,
"แŒฆ": 112,
"แŒง": 95,
"แŒจ": 42,
"แŒฉ": 63,
"แŒช": 177,
"แŒซ": 106,
"แŒฌ": 186,
"แŒญ": 99,
"แŒฎ": 122,
"แŒฒ": 195,
"แŒด": 73,
"แŒต": 144,
"แŒถ": 172,
"แŒธ": 166,
"แŒน": 21,
"แŒบ": 115,
"แŒป": 111,
"แŒฝ": 33,
"แŒพ": 211,
"แŒฟ": 77,
"แ€": 218,
"แ": 163,
"แƒ": 94,
"แ…": 199,
"แˆ": 210,
"แ‰": 118,
"แŠ": 84,
"แ‹": 38,
"แŒ": 220,
"แ": 119,
"แŽ": 60,
"แ": 72,
"แ‘": 18,
"แ’": 148,
"แ“": 4,
"แ”": 208,
"แ•": 82,
"แ–": 76,
"โ€œ": 87,
"โ€": 9
}