Upload tokenizer

vocab.json

@@ -1,232 +1,232 @@
 {
   "[PAD]": 229,
   "[UNK]": 228,
-  "|": 161,
-  "ሀ": 219,
-  "ሁ": 162,
-  "ሂ": 67,
-  "ሃ": 191,
-  "ሄ": 125,
-  "ህ": 80,
-  "ሆ": 184,
-  "ለ": 53,
-  "ሉ": 109,
-  "ሊ": 103,
-  "ላ": 36,
-  "ሌ": 183,
-  "ል": 46,
-  "ሎ": 139,
-  "ሏ": 17,
-  "ሐ": 1,
-  "ሑ": 130,
-  "ሓ": 227,
-  "ሔ": 16,
-  "ሕ": 209,
-  "መ": 200,
-  "ሙ": 11,
-  "ሚ": 85,
-  "ማ": 152,
-  "ሜ": 35,
-  "ም": 135,
-  "ሞ": 173,
-  "ሟ": 143,
-  "ሠ": 165,
-  "ሡ": 12,
-  "ሣ": 22,
-  "ሥ": 91,
-  "ሦ": 97,
-  "ረ": 30,
-  "ሩ": 19,
-  "ሪ": 29,
-  "ራ": 79,
-  "ሬ": 0,
-  "ር": 101,
-  "ሮ": 142,
-  "ሯ": 27,
-  "ሰ": 26,
-  "ሱ": 132,
-  "ሲ": 20,
-  "ሳ": 107,
-  "ሴ": 45,
-  "ስ": 178,
-  "ሶ": 167,
-  "ሷ": 43,
-  "ሸ": 48,
-  "ሹ": 86,
-  "ሺ": 62,
-  "ሻ": 24,
-  "ሼ": 83,
-  "ሽ": 55,
-  "ሾ": 168,
-  "ቀ": 116,
-  "ቁ": 75,
-  "ቂ": 214,
-  "ቃ": 192,
-  "ቄ": 126,
-  "ቅ": 5,
-  "ቆ": 215,
-  "ቋ": 28,
-  "በ": 188,
-  "ቡ": 98,
-  "ቢ": 13,
-  "ባ": 50,
-  "ቤ": 3,
-  "ብ": 39,
-  "ቦ": 151,
-  "ቧ": 124,
-  "ቨ": 34,
-  "ቪ": 181,
-  "ቫ": 68,
-  "ቭ": 102,
-  "ቮ": 194,
-  "ተ": 96,
-  "ቱ": 6,
-  "ቲ": 61,
-  "ታ": 74,
-  "ቴ": 153,
-  "ት": 158,
-  "ቶ": 222,
-  "ቷ": 185,
-  "ቸ": 41,
-  "ቹ": 207,
-  "ቺ": 15,
-  "ቻ": 52,
-  "ቼ": 224,
-  "ች": 145,
-  "ቾ": 51,
-  "ቿ": 37,
-  "ኀ": 147,
-  "ኃ": 160,
-  "ኅ": 197,
-  "ኋ": 58,
-  "ነ": 129,
-  "ኑ": 204,
-  "ኒ": 44,
-  "ና": 104,
-  "ኔ": 93,
-  "ን": 180,
-  "ኖ": 221,
-  "ኗ": 88,
-  "ኘ": 174,
-  "ኙ": 100,
-  "ኛ": 226,
-  "ኝ": 176,
-  "ኞ": 40,
-  "ኟ": 89,
-  "አ": 141,
-  "ኡ": 206,
-  "ኢ": 31,
-  "ኤ": 70,
-  "እ": 179,
-  "ኦ": 66,
-  "ከ": 8,
-  "ኩ": 205,
-  "ኪ": 32,
-  "ካ": 123,
-  "ኬ": 170,
-  "ክ": 156,
-  "ኮ": 127,
-  "ኳ": 150,
-  "ኸ": 225,
-  "ኽ": 189,
-  "ወ": 198,
-  "ዊ": 187,
-  "ዋ": 154,
-  "ዌ": 69,
-  "ው": 164,
-  "ዎ": 108,
-  "ዐ": 117,
-  "ዑ": 216,
-  "ዒ": 54,
-  "ዓ": 56,
-  "ዕ": 105,
-  "ዖ": 47,
-  "ዘ": 90,
-  "ዙ": 202,
-  "ዚ": 140,
-  "ዛ": 212,
-  "ዜ": 113,
-  "ዝ": 171,
-  "ዞ": 81,
-  "ዟ": 155,
-  "ዢ": 201,
-  "ዣ": 131,
-  "ዤ": 120,
-  "ዥ": 23,
-  "ዦ": 159,
-  "የ": 134,
-  "ዩ": 14,
-  "ያ": 182,
-  "ዬ": 138,
-  "ይ": 213,
-  "ዮ": 78,
-  "ደ": 10,
-  "ዱ": 128,
-  "ዲ": 121,
-  "ዳ": 190,
-  "ዴ": 196,
-  "ድ": 169,
-  "ዶ": 49,
-  "ዷ": 59,
-  "ጀ": 157,
-  "ጁ": 149,
-  "ጂ": 203,
-  "ጃ": 136,
-  "ጄ": 7,
-  "ጅ": 57,
-  "ጆ": 114,
-  "ገ": 137,
-  "ጉ": 146,
-  "ጊ": 65,
-  "ጋ": 217,
-  "ጌ": 223,
-  "ግ": 25,
-  "ጎ": 110,
-  "ጓ": 64,
-  "ጠ": 175,
-  "ጡ": 193,
-  "ጢ": 133,
-  "ጣ": 92,
-  "ጤ": 71,
-  "ጥ": 2,
-  "ጦ": 112,
-  "ጧ": 95,
-  "ጨ": 42,
-  "ጩ": 63,
-  "ጪ": 177,
-  "ጫ": 106,
-  "ጬ": 186,
-  "ጭ": 99,
-  "ጮ": 122,
-  "ጲ": 195,
-  "ጴ": 73,
-  "ጵ": 144,
-  "ጶ": 172,
-  "ጸ": 166,
-  "ጹ": 21,
-  "ጺ": 115,
-  "ጻ": 111,
-  "ጽ": 33,
-  "ጾ": 211,
-  "ጿ": 77,
-  "ፀ": 218,
-  "ፁ": 163,
-  "ፃ": 94,
-  "ፅ": 199,
-  "ፈ": 210,
-  "ፉ": 118,
-  "ፊ": 84,
-  "ፋ": 38,
-  "ፌ": 220,
-  "ፍ": 119,
-  "ፎ": 60,
-  "ፏ": 72,
-  "ፑ": 18,
-  "ፒ": 148,
-  "ፓ": 4,
-  "ፔ": 208,
-  "ፕ": 82,
-  "ፖ": 76,
-  "“": 87,
-  "”": 9
+  "|": 114,
+  "ሀ": 225,
+  "ሁ": 43,
+  "ሂ": 24,
+  "ሃ": 189,
+  "ሄ": 32,
+  "ህ": 129,
+  "ሆ": 221,
+  "ለ": 152,
+  "ሉ": 108,
+  "ሊ": 69,
+  "ላ": 4,
+  "ሌ": 102,
+  "ል": 17,
+  "ሎ": 50,
+  "ሏ": 30,
+  "ሐ": 60,
+  "ሑ": 35,
+  "ሓ": 206,
+  "ሔ": 82,
+  "ሕ": 128,
+  "መ": 56,
+  "ሙ": 210,
+  "ሚ": 41,
+  "ማ": 166,
+  "ሜ": 180,
+  "ም": 40,
+  "ሞ": 226,
+  "ሟ": 194,
+  "ሠ": 47,
+  "ሡ": 149,
+  "ሣ": 86,
+  "ሥ": 99,
+  "ሦ": 81,
+  "ረ": 88,
+  "ሩ": 190,
+  "ሪ": 185,
+  "ራ": 135,
+  "ሬ": 162,
+  "ር": 58,
+  "ሮ": 48,
+  "ሯ": 39,
+  "ሰ": 105,
+  "ሱ": 148,
+  "ሲ": 144,
+  "ሳ": 156,
+  "ሴ": 95,
+  "ስ": 27,
+  "ሶ": 115,
+  "ሷ": 53,
+  "ሸ": 94,
+  "ሹ": 85,
+  "ሺ": 215,
+  "ሻ": 3,
+  "ሼ": 179,
+  "ሽ": 193,
+  "ሾ": 71,
+  "ቀ": 200,
+  "ቁ": 224,
+  "ቂ": 87,
+  "ቃ": 160,
+  "ቄ": 153,
+  "ቅ": 163,
+  "ቆ": 175,
+  "ቋ": 204,
+  "በ": 101,
+  "ቡ": 34,
+  "ቢ": 170,
+  "ባ": 122,
+  "ቤ": 109,
+  "ብ": 79,
+  "ቦ": 171,
+  "ቧ": 46,
+  "ቨ": 61,
+  "ቪ": 137,
+  "ቫ": 143,
+  "ቭ": 10,
+  "ቮ": 74,
+  "ተ": 165,
+  "ቱ": 51,
+  "ቲ": 49,
+  "ታ": 130,
+  "ቴ": 184,
+  "ት": 107,
+  "ቶ": 191,
+  "ቷ": 178,
+  "ቸ": 63,
+  "ቹ": 223,
+  "ቺ": 209,
+  "ቻ": 25,
+  "ቼ": 116,
+  "ች": 5,
+  "ቾ": 77,
+  "ቿ": 158,
+  "ኀ": 1,
+  "ኃ": 119,
+  "ኅ": 222,
+  "ኋ": 205,
+  "ነ": 14,
+  "ኑ": 8,
+  "ኒ": 211,
+  "ና": 83,
+  "ኔ": 151,
+  "ን": 172,
+  "ኖ": 201,
+  "ኗ": 216,
+  "ኘ": 33,
+  "ኙ": 123,
+  "ኛ": 169,
+  "ኝ": 183,
+  "ኞ": 97,
+  "ኟ": 177,
+  "አ": 45,
+  "ኡ": 145,
+  "ኢ": 154,
+  "ኤ": 13,
+  "እ": 75,
+  "ኦ": 12,
+  "ከ": 140,
+  "ኩ": 64,
+  "ኪ": 113,
+  "ካ": 36,
+  "ኬ": 134,
+  "ክ": 192,
+  "ኮ": 168,
+  "ኳ": 125,
+  "ኸ": 93,
+  "ኽ": 197,
+  "ወ": 72,
+  "ዊ": 142,
+  "ዋ": 65,
+  "ዌ": 187,
+  "ው": 198,
+  "ዎ": 213,
+  "ዐ": 188,
+  "ዑ": 212,
+  "ዒ": 7,
+  "ዓ": 139,
+  "ዕ": 132,
+  "ዖ": 124,
+  "ዘ": 117,
+  "ዙ": 100,
+  "ዚ": 121,
+  "ዛ": 181,
+  "ዜ": 173,
+  "ዝ": 15,
+  "ዞ": 16,
+  "ዟ": 161,
+  "ዢ": 167,
+  "ዣ": 217,
+  "ዤ": 159,
+  "ዥ": 186,
+  "ዦ": 126,
+  "የ": 0,
+  "ዩ": 67,
+  "ያ": 76,
+  "ዬ": 104,
+  "ይ": 37,
+  "ዮ": 2,
+  "ደ": 202,
+  "ዱ": 220,
+  "ዲ": 155,
+  "ዳ": 28,
+  "ዴ": 23,
+  "ድ": 31,
+  "ዶ": 120,
+  "ዷ": 92,
+  "ጀ": 29,
+  "ጁ": 207,
+  "ጂ": 57,
+  "ጃ": 208,
+  "ጄ": 203,
+  "ጅ": 103,
+  "ጆ": 9,
+  "ገ": 42,
+  "ጉ": 68,
+  "ጊ": 84,
+  "ጋ": 141,
+  "ጌ": 146,
+  "ግ": 118,
+  "ጎ": 22,
+  "ጓ": 90,
+  "ጠ": 150,
+  "ጡ": 62,
+  "ጢ": 131,
+  "ጣ": 19,
+  "ጤ": 20,
+  "ጥ": 106,
+  "ጦ": 11,
+  "ጧ": 52,
+  "ጨ": 199,
+  "ጩ": 55,
+  "ጪ": 70,
+  "ጫ": 73,
+  "ጬ": 110,
+  "ጭ": 127,
+  "ጮ": 91,
+  "ጲ": 196,
+  "ጴ": 138,
+  "ጵ": 219,
+  "ጶ": 111,
+  "ጸ": 59,
+  "ጹ": 98,
+  "ጺ": 195,
+  "ጻ": 96,
+  "ጽ": 21,
+  "ጾ": 80,
+  "ጿ": 26,
+  "ፀ": 164,
+  "ፁ": 133,
+  "ፃ": 38,
+  "ፅ": 18,
+  "ፈ": 6,
+  "ፉ": 218,
+  "ፊ": 89,
+  "ፋ": 112,
+  "ፌ": 176,
+  "ፍ": 54,
+  "ፎ": 136,
+  "ፏ": 227,
+  "ፑ": 66,
+  "ፒ": 78,
+  "ፓ": 182,
+  "ፔ": 147,
+  "ፕ": 174,
+  "ፖ": 214,
+  "“": 157,
+  "”": 44
 }