mms-tts-khm / vocab.json
sanchit-gandhi's picture
Upload tokenizer
e757558
raw
history blame contribute delete
945 Bytes
{
" ": 19,
"-": 71,
"1": 57,
"_": 36,
"q": 15,
"แž€": 38,
"แž": 62,
"แž‚": 4,
"แžƒ": 9,
"แž„": 5,
"แž…": 54,
"แž†": 3,
"แž‡": 20,
"แžˆ": 14,
"แž‰": 64,
"แžŠ": 42,
"แž‹": 55,
"แžŒ": 10,
"แžŽ": 56,
"แž": 35,
"แž": 68,
"แž‘": 24,
"แž’": 7,
"แž“": 59,
"แž”": 12,
"แž•": 17,
"แž–": 52,
"แž—": 73,
"แž˜": 72,
"แž™": 1,
"แžš": 26,
"แž›": 22,
"แžœ": 43,
"แžŸ": 61,
"แž ": 58,
"แžก": 53,
"แžข": 2,
"แžฅ": 11,
"แžง": 21,
"แžช": 70,
"แžซ": 46,
"แžฌ": 45,
"แžญ": 28,
"แžฎ": 66,
"แžฏ": 51,
"แžฑ": 23,
"แžถ": 29,
"แžท": 18,
"แžธ": 49,
"แžน": 60,
"แžบ": 69,
"แžป": 50,
"แžผ": 25,
"แžฝ": 13,
"แžพ": 6,
"แžฟ": 31,
"แŸ€": 30,
"แŸ": 32,
"แŸ‚": 39,
"แŸƒ": 41,
"แŸ„": 48,
"แŸ…": 40,
"แŸ†": 33,
"แŸ‡": 47,
"แŸˆ": 44,
"แŸ‰": 37,
"แŸŠ": 65,
"แŸ‹": 8,
"แŸŒ": 67,
"แŸ": 34,
"แŸŽ": 27,
"แŸ": 16,
"แŸ": 63,
"แŸ’": 0
}