Thunpitcha's picture
Upload tokenizer
2c8ffcd
raw
history blame
1.06 kB
{
"'": 57,
"[PAD]": 83,
"[UNK]": 82,
"_": 65,
"a": 59,
"c": 20,
"e": 31,
"h": 25,
"i": 18,
"j": 71,
"n": 12,
"o": 44,
"s": 69,
"t": 53,
"|": 72,
"~": 46,
"ก": 60,
"ข": 64,
"ค": 41,
"ฆ": 52,
"ง": 42,
"จ": 66,
"ฉ": 70,
"ช": 9,
"ซ": 11,
"ฌ": 15,
"ญ": 32,
"ฎ": 37,
"ฏ": 28,
"ฐ": 49,
"ฑ": 33,
"ฒ": 61,
"ณ": 81,
"ด": 56,
"ต": 39,
"ถ": 58,
"ท": 76,
"ธ": 26,
"น": 35,
"บ": 48,
"ป": 13,
"ผ": 36,
"ฝ": 16,
"พ": 6,
"ฟ": 40,
"ภ": 24,
"ม": 34,
"ย": 7,
"ร": 45,
"ฤ": 19,
"ล": 10,
"ว": 67,
"ศ": 14,
"ษ": 79,
"ส": 29,
"ห": 8,
"ฬ": 2,
"อ": 68,
"ฮ": 47,
"ะ": 51,
"ั": 50,
"า": 63,
"ำ": 5,
"ิ": 0,
"ี": 3,
"ึ": 62,
"ื": 27,
"ุ": 1,
"ู": 17,
"เ": 21,
"แ": 43,
"โ": 80,
"ใ": 54,
"ไ": 73,
"ๅ": 55,
"ๆ": 22,
"็": 74,
"่": 23,
"้": 77,
"๊": 78,
"๋": 75,
"์": 30,
"ํ": 38,
"’": 4
}