Thunpitcha's picture
Upload tokenizer
1130606
raw
history blame
877 Bytes
{
"'": 12,
"[PAD]": 67,
"[UNK]": 66,
"|": 31,
"ก": 48,
"ข": 37,
"ค": 6,
"ฆ": 65,
"ง": 45,
"จ": 46,
"ฉ": 64,
"ช": 20,
"ซ": 61,
"ญ": 10,
"ฎ": 43,
"ฏ": 25,
"ฐ": 56,
"ฑ": 55,
"ฒ": 4,
"ณ": 40,
"ด": 47,
"ต": 21,
"ถ": 62,
"ท": 9,
"ธ": 59,
"น": 30,
"บ": 41,
"ป": 2,
"ผ": 17,
"ฝ": 52,
"พ": 32,
"ฟ": 13,
"ภ": 27,
"ม": 49,
"ย": 33,
"ร": 53,
"ฤ": 44,
"ล": 7,
"ว": 38,
"ศ": 39,
"ษ": 63,
"ส": 23,
"ห": 36,
"ฬ": 3,
"อ": 5,
"ฮ": 51,
"ะ": 34,
"ั": 60,
"า": 0,
"ำ": 16,
"ิ": 15,
"ี": 24,
"ึ": 54,
"ื": 14,
"ุ": 22,
"ู": 42,
"เ": 50,
"แ": 28,
"โ": 11,
"ใ": 19,
"ไ": 18,
"ๆ": 26,
"็": 29,
"่": 58,
"้": 1,
"๊": 8,
"๋": 35,
"์": 57
}