Upload tokenizer

vocab.json

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