ndeclarke's picture
Upload tokenizer
84a8f84 verified
raw
history blame
2.25 kB
{
"[PAD]": 167,
"[UNK]": 166,
"|": 0,
"ሀ": 1,
"ሁ": 2,
"ሂ": 3,
"ሃ": 4,
"ሄ": 5,
"ህ": 6,
"ሆ": 7,
"ለ": 8,
"ሉ": 9,
"ሊ": 10,
"ላ": 11,
"ሌ": 12,
"ል": 13,
"ሎ": 14,
"ሐ": 15,
"ሑ": 16,
"ሒ": 17,
"ሓ": 18,
"ሔ": 19,
"ሕ": 20,
"ሖ": 21,
"መ": 22,
"ሙ": 23,
"ሚ": 24,
"ማ": 25,
"ሜ": 26,
"ም": 27,
"ሞ": 28,
"ረ": 29,
"ሩ": 30,
"ሪ": 31,
"ራ": 32,
"ሬ": 33,
"ር": 34,
"ሮ": 35,
"ሰ": 36,
"ሱ": 37,
"ሲ": 38,
"ሳ": 39,
"ሴ": 40,
"ስ": 41,
"ሶ": 42,
"ሸ": 43,
"ሹ": 44,
"ሺ": 45,
"ሻ": 46,
"ሼ": 47,
"ሽ": 48,
"ሾ": 49,
"ቀ": 50,
"ቁ": 51,
"ቂ": 52,
"ቃ": 53,
"ቄ": 54,
"ቅ": 55,
"ቆ": 56,
"በ": 57,
"ቡ": 58,
"ቢ": 59,
"ባ": 60,
"ቤ": 61,
"ብ": 62,
"ቦ": 63,
"ተ": 64,
"ቱ": 65,
"ቲ": 66,
"ታ": 67,
"ቴ": 68,
"ት": 69,
"ቶ": 70,
"ቹ": 71,
"ቼ": 72,
"ነ": 73,
"ኑ": 74,
"ኒ": 75,
"ና": 76,
"ኔ": 77,
"ን": 78,
"ኖ": 79,
"አ": 80,
"ኡ": 81,
"ኢ": 82,
"ኣ": 83,
"ኤ": 84,
"እ": 85,
"ኦ": 86,
"ከ": 87,
"ኩ": 88,
"ኪ": 89,
"ካ": 90,
"ኬ": 91,
"ክ": 92,
"ኮ": 93,
"ወ": 94,
"ዉ": 95,
"ዊ": 96,
"ዋ": 97,
"ዌ": 98,
"ው": 99,
"ዎ": 100,
"ዐ": 101,
"ዑ": 102,
"ዒ": 103,
"ዓ": 104,
"ዔ": 105,
"ዕ": 106,
"ዖ": 107,
"ዘ": 108,
"ዙ": 109,
"ዚ": 110,
"ዛ": 111,
"ዜ": 112,
"ዝ": 113,
"ዞ": 114,
"የ": 115,
"ዩ": 116,
"ያ": 117,
"ዬ": 118,
"ይ": 119,
"ዮ": 120,
"ደ": 121,
"ዱ": 122,
"ዲ": 123,
"ዳ": 124,
"ዴ": 125,
"ድ": 126,
"ዶ": 127,
"ጀ": 128,
"ጁ": 129,
"ጂ": 130,
"ጃ": 131,
"ጄ": 132,
"ጅ": 133,
"ጆ": 134,
"ገ": 135,
"ጉ": 136,
"ጊ": 137,
"ጋ": 138,
"ጌ": 139,
"ግ": 140,
"ጎ": 141,
"ጠ": 142,
"ጡ": 143,
"ጢ": 144,
"ጣ": 145,
"ጥ": 146,
"ጦ": 147,
"ጨ": 148,
"ጪ": 149,
"ጫ": 150,
"ጭ": 151,
"ጸ": 152,
"ጹ": 153,
"ጺ": 154,
"ጻ": 155,
"ጽ": 156,
"ጾ": 157,
"ፈ": 158,
"ፉ": 159,
"ፊ": 160,
"ፋ": 161,
"ፌ": 162,
"ፍ": 163,
"ፎ": 164,
"’": 165
}