timotewos's picture
Upload tokenizer
9e221ce verified
raw
history blame
3.12 kB
{
"[PAD]": 229,
"[UNK]": 228,
"|": 114,
"แˆ€": 225,
"แˆ": 43,
"แˆ‚": 24,
"แˆƒ": 189,
"แˆ„": 32,
"แˆ…": 129,
"แˆ†": 221,
"แˆˆ": 152,
"แˆ‰": 108,
"แˆŠ": 69,
"แˆ‹": 4,
"แˆŒ": 102,
"แˆ": 17,
"แˆŽ": 50,
"แˆ": 30,
"แˆ": 60,
"แˆ‘": 35,
"แˆ“": 206,
"แˆ”": 82,
"แˆ•": 128,
"แˆ˜": 56,
"แˆ™": 210,
"แˆš": 41,
"แˆ›": 166,
"แˆœ": 180,
"แˆ": 40,
"แˆž": 226,
"แˆŸ": 194,
"แˆ ": 47,
"แˆก": 149,
"แˆฃ": 86,
"แˆฅ": 99,
"แˆฆ": 81,
"แˆจ": 88,
"แˆฉ": 190,
"แˆช": 185,
"แˆซ": 135,
"แˆฌ": 162,
"แˆญ": 58,
"แˆฎ": 48,
"แˆฏ": 39,
"แˆฐ": 105,
"แˆฑ": 148,
"แˆฒ": 144,
"แˆณ": 156,
"แˆด": 95,
"แˆต": 27,
"แˆถ": 115,
"แˆท": 53,
"แˆธ": 94,
"แˆน": 85,
"แˆบ": 215,
"แˆป": 3,
"แˆผ": 179,
"แˆฝ": 193,
"แˆพ": 71,
"แ‰€": 200,
"แ‰": 224,
"แ‰‚": 87,
"แ‰ƒ": 160,
"แ‰„": 153,
"แ‰…": 163,
"แ‰†": 175,
"แ‰‹": 204,
"แ‰ ": 101,
"แ‰ก": 34,
"แ‰ข": 170,
"แ‰ฃ": 122,
"แ‰ค": 109,
"แ‰ฅ": 79,
"แ‰ฆ": 171,
"แ‰ง": 46,
"แ‰จ": 61,
"แ‰ช": 137,
"แ‰ซ": 143,
"แ‰ญ": 10,
"แ‰ฎ": 74,
"แ‰ฐ": 165,
"แ‰ฑ": 51,
"แ‰ฒ": 49,
"แ‰ณ": 130,
"แ‰ด": 184,
"แ‰ต": 107,
"แ‰ถ": 191,
"แ‰ท": 178,
"แ‰ธ": 63,
"แ‰น": 223,
"แ‰บ": 209,
"แ‰ป": 25,
"แ‰ผ": 116,
"แ‰ฝ": 5,
"แ‰พ": 77,
"แ‰ฟ": 158,
"แŠ€": 1,
"แŠƒ": 119,
"แŠ…": 222,
"แŠ‹": 205,
"แŠ": 14,
"แŠ‘": 8,
"แŠ’": 211,
"แŠ“": 83,
"แŠ”": 151,
"แŠ•": 172,
"แŠ–": 201,
"แŠ—": 216,
"แŠ˜": 33,
"แŠ™": 123,
"แŠ›": 169,
"แŠ": 183,
"แŠž": 97,
"แŠŸ": 177,
"แŠ ": 45,
"แŠก": 145,
"แŠข": 154,
"แŠค": 13,
"แŠฅ": 75,
"แŠฆ": 12,
"แŠจ": 140,
"แŠฉ": 64,
"แŠช": 113,
"แŠซ": 36,
"แŠฌ": 134,
"แŠญ": 192,
"แŠฎ": 168,
"แŠณ": 125,
"แŠธ": 93,
"แŠฝ": 197,
"แ‹ˆ": 72,
"แ‹Š": 142,
"แ‹‹": 65,
"แ‹Œ": 187,
"แ‹": 198,
"แ‹Ž": 213,
"แ‹": 188,
"แ‹‘": 212,
"แ‹’": 7,
"แ‹“": 139,
"แ‹•": 132,
"แ‹–": 124,
"แ‹˜": 117,
"แ‹™": 100,
"แ‹š": 121,
"แ‹›": 181,
"แ‹œ": 173,
"แ‹": 15,
"แ‹ž": 16,
"แ‹Ÿ": 161,
"แ‹ข": 167,
"แ‹ฃ": 217,
"แ‹ค": 159,
"แ‹ฅ": 186,
"แ‹ฆ": 126,
"แ‹จ": 0,
"แ‹ฉ": 67,
"แ‹ซ": 76,
"แ‹ฌ": 104,
"แ‹ญ": 37,
"แ‹ฎ": 2,
"แ‹ฐ": 202,
"แ‹ฑ": 220,
"แ‹ฒ": 155,
"แ‹ณ": 28,
"แ‹ด": 23,
"แ‹ต": 31,
"แ‹ถ": 120,
"แ‹ท": 92,
"แŒ€": 29,
"แŒ": 207,
"แŒ‚": 57,
"แŒƒ": 208,
"แŒ„": 203,
"แŒ…": 103,
"แŒ†": 9,
"แŒˆ": 42,
"แŒ‰": 68,
"แŒŠ": 84,
"แŒ‹": 141,
"แŒŒ": 146,
"แŒ": 118,
"แŒŽ": 22,
"แŒ“": 90,
"แŒ ": 150,
"แŒก": 62,
"แŒข": 131,
"แŒฃ": 19,
"แŒค": 20,
"แŒฅ": 106,
"แŒฆ": 11,
"แŒง": 52,
"แŒจ": 199,
"แŒฉ": 55,
"แŒช": 70,
"แŒซ": 73,
"แŒฌ": 110,
"แŒญ": 127,
"แŒฎ": 91,
"แŒฒ": 196,
"แŒด": 138,
"แŒต": 219,
"แŒถ": 111,
"แŒธ": 59,
"แŒน": 98,
"แŒบ": 195,
"แŒป": 96,
"แŒฝ": 21,
"แŒพ": 80,
"แŒฟ": 26,
"แ€": 164,
"แ": 133,
"แƒ": 38,
"แ…": 18,
"แˆ": 6,
"แ‰": 218,
"แŠ": 89,
"แ‹": 112,
"แŒ": 176,
"แ": 54,
"แŽ": 136,
"แ": 227,
"แ‘": 66,
"แ’": 78,
"แ“": 182,
"แ”": 147,
"แ•": 174,
"แ–": 214,
"โ€œ": 157,
"โ€": 44
}