tamasheq-99 / vocab.json
ad019el's picture
Upload tokenizer
b478b17
raw
history blame
1.18 kB
{
":": 59,
"J": 12,
"[PAD]": 96,
"[UNK]": 95,
"a": 74,
"b": 45,
"c": 60,
"d": 5,
"e": 46,
"f": 56,
"g": 84,
"h": 29,
"i": 72,
"j": 70,
"k": 82,
"l": 67,
"m": 4,
"n": 17,
"o": 94,
"p": 50,
"q": 65,
"r": 90,
"s": 68,
"t": 24,
"u": 23,
"v": 14,
"w": 37,
"x": 49,
"y": 25,
"z": 73,
"|": 43,
"°": 38,
"à": 13,
"ä": 31,
"å": 86,
"è": 78,
"ì": 62,
"ò": 1,
"ö": 6,
"ù": 81,
"ą": 66,
"č": 89,
"ė": 54,
"ę": 44,
"į": 0,
"š": 51,
"ū": 47,
"ų": 3,
"ž": 61,
"அ": 42,
"ஆ": 27,
"இ": 52,
"ஈ": 30,
"உ": 77,
"ஊ": 8,
"எ": 35,
"ஏ": 85,
"ஐ": 41,
"ஒ": 91,
"ஓ": 48,
"க": 34,
"ங": 93,
"ச": 21,
"ஜ": 40,
"ஞ": 57,
"ட": 88,
"ண": 15,
"த": 64,
"ந": 26,
"ன": 18,
"ப": 11,
"ம": 53,
"ய": 83,
"ர": 2,
"ற": 76,
"ல": 39,
"ள": 55,
"ழ": 92,
"வ": 63,
"ஷ": 28,
"ஸ": 19,
"ஹ": 20,
"ா": 32,
"ி": 9,
"ீ": 58,
"ு": 87,
"ூ": 69,
"ெ": 16,
"ே": 7,
"ை": 36,
"ொ": 10,
"ோ": 75,
"ௌ": 33,
"்": 80,
"’": 79,
"„": 22,
"…": 71
}