Upload tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
   "</s>": 231,
-  "[PAD]": 230
+  "<s>": 230
 }

tokenizer_config.json

@@ -1,6 +1,14 @@
 {
   "added_tokens_decoder": {
-    "230": {
+    "228": {
+      "content": "[UNK]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "229": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -8,6 +16,14 @@
       "single_word": false,
       "special": false
     },
+    "230": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
     "231": {
       "content": "</s>",
       "lstrip": false,

vocab.json

@@ -1,232 +1,232 @@
 {
-  "[PAD]": 230,
-  "[UNK]": 230,
-  "|": 24,
-  "ሀ": 12,
-  "ሁ": 38,
-  "ሂ": 197,
-  "ሃ": 173,
-  "ሄ": 128,
-  "ህ": 98,
-  "ሆ": 3,
-  "ለ": 4,
-  "ሉ": 224,
-  "ሊ": 181,
-  "ላ": 163,
-  "ሌ": 211,
-  "ል": 96,
-  "ሎ": 167,
-  "ሏ": 103,
-  "ሐ": 85,
-  "ሑ": 62,
-  "ሓ": 82,
-  "ሔ": 223,
-  "ሕ": 161,
-  "መ": 30,
-  "ሙ": 58,
-  "ሚ": 182,
-  "ማ": 188,
-  "ሜ": 194,
-  "ም": 41,
-  "ሞ": 185,
-  "ሟ": 143,
-  "ሠ": 205,
-  "ሡ": 151,
-  "ሣ": 171,
-  "ሥ": 72,
-  "ሦ": 70,
-  "ረ": 123,
-  "ሩ": 104,
-  "ሪ": 206,
-  "ራ": 152,
-  "ሬ": 105,
-  "ር": 73,
-  "ሮ": 43,
-  "ሯ": 226,
-  "ሰ": 156,
-  "ሱ": 121,
-  "ሲ": 39,
-  "ሳ": 64,
-  "ሴ": 9,
-  "ስ": 145,
-  "ሶ": 180,
-  "ሷ": 192,
-  "ሸ": 8,
-  "ሹ": 183,
-  "ሺ": 27,
-  "ሻ": 63,
-  "ሼ": 150,
-  "ሽ": 32,
-  "ሾ": 76,
-  "ቀ": 222,
-  "ቁ": 2,
-  "ቂ": 141,
-  "ቃ": 129,
-  "ቄ": 187,
-  "ቅ": 120,
-  "ቆ": 91,
-  "ቋ": 202,
-  "በ": 90,
-  "ቡ": 175,
-  "ቢ": 46,
-  "ባ": 136,
-  "ቤ": 190,
-  "ብ": 80,
-  "ቦ": 174,
-  "ቧ": 138,
-  "ቨ": 172,
-  "ቪ": 15,
-  "ቫ": 6,
-  "ቭ": 155,
-  "ቮ": 34,
-  "ተ": 26,
-  "ቱ": 164,
-  "ቲ": 144,
-  "ታ": 78,
-  "ቴ": 93,
-  "ት": 61,
-  "ቶ": 148,
-  "ቷ": 25,
-  "ቸ": 68,
-  "ቹ": 10,
-  "ቺ": 86,
-  "ቻ": 110,
-  "ቼ": 195,
-  "ች": 19,
-  "ቾ": 74,
-  "ቿ": 83,
-  "ኀ": 139,
-  "ኃ": 159,
-  "ኅ": 65,
-  "ኋ": 13,
-  "ነ": 107,
-  "ኑ": 184,
-  "ኒ": 77,
-  "ና": 218,
-  "ኔ": 210,
-  "ን": 108,
-  "ኖ": 204,
-  "ኗ": 51,
-  "ኘ": 48,
-  "ኙ": 52,
-  "ኛ": 119,
-  "ኝ": 45,
-  "ኞ": 201,
-  "ኟ": 177,
-  "አ": 92,
-  "ኡ": 142,
-  "ኢ": 133,
-  "ኤ": 115,
-  "እ": 140,
-  "ኦ": 36,
-  "ከ": 17,
-  "ኩ": 131,
-  "ኪ": 102,
-  "ካ": 178,
-  "ኬ": 57,
-  "ክ": 135,
-  "ኮ": 54,
-  "ኳ": 23,
-  "ኸ": 21,
-  "ኽ": 209,
-  "ወ": 168,
-  "ዊ": 153,
-  "ዋ": 165,
-  "ዌ": 169,
-  "ው": 191,
-  "ዎ": 198,
-  "ዐ": 146,
-  "ዑ": 95,
-  "ዒ": 18,
-  "ዓ": 217,
-  "ዕ": 125,
-  "ዖ": 67,
-  "ዘ": 132,
-  "ዙ": 227,
-  "ዚ": 220,
-  "ዛ": 126,
-  "ዜ": 130,
-  "ዝ": 117,
-  "ዞ": 118,
-  "ዟ": 157,
-  "ዢ": 154,
-  "ዣ": 106,
-  "ዤ": 147,
-  "ዥ": 79,
-  "ዦ": 7,
-  "የ": 200,
-  "ዩ": 71,
-  "ያ": 166,
-  "ዬ": 127,
-  "ይ": 179,
-  "ዮ": 87,
-  "ደ": 170,
-  "ዱ": 89,
-  "ዲ": 0,
-  "ዳ": 122,
-  "ዴ": 42,
-  "ድ": 134,
-  "ዶ": 207,
-  "ዷ": 214,
-  "ጀ": 14,
+  "[PAD]": 229,
+  "[UNK]": 228,
+  "|": 144,
+  "ሀ": 78,
+  "ሁ": 165,
+  "ሂ": 120,
+  "ሃ": 77,
+  "ሄ": 200,
+  "ህ": 142,
+  "ሆ": 46,
+  "ለ": 150,
+  "ሉ": 44,
+  "ሊ": 79,
+  "ላ": 126,
+  "ሌ": 8,
+  "ል": 58,
+  "ሎ": 40,
+  "ሏ": 194,
+  "ሐ": 134,
+  "ሑ": 167,
+  "ሓ": 164,
+  "ሔ": 110,
+  "ሕ": 102,
+  "መ": 184,
+  "ሙ": 9,
+  "ሚ": 87,
+  "ማ": 148,
+  "ሜ": 24,
+  "ም": 7,
+  "ሞ": 172,
+  "ሟ": 218,
+  "ሠ": 220,
+  "ሡ": 176,
+  "ሣ": 141,
+  "ሥ": 27,
+  "ሦ": 192,
+  "ረ": 55,
+  "ሩ": 170,
+  "ሪ": 107,
+  "ራ": 22,
+  "ሬ": 5,
+  "ር": 136,
+  "ሮ": 81,
+  "ሯ": 157,
+  "ሰ": 94,
+  "ሱ": 13,
+  "ሲ": 1,
+  "ሳ": 2,
+  "ሴ": 208,
+  "ስ": 191,
+  "ሶ": 80,
+  "ሷ": 123,
+  "ሸ": 175,
+  "ሹ": 34,
+  "ሺ": 129,
+  "ሻ": 226,
+  "ሼ": 223,
+  "ሽ": 187,
+  "ሾ": 163,
+  "ቀ": 198,
+  "ቁ": 90,
+  "ቂ": 35,
+  "ቃ": 207,
+  "ቄ": 60,
+  "ቅ": 173,
+  "ቆ": 222,
+  "ቋ": 195,
+  "በ": 169,
+  "ቡ": 75,
+  "ቢ": 97,
+  "ባ": 186,
+  "ቤ": 202,
+  "ብ": 151,
+  "ቦ": 188,
+  "ቧ": 182,
+  "ቨ": 158,
+  "ቪ": 95,
+  "ቫ": 36,
+  "ቭ": 156,
+  "ቮ": 197,
+  "ተ": 114,
+  "ቱ": 43,
+  "ቲ": 139,
+  "ታ": 23,
+  "ቴ": 105,
+  "ት": 138,
+  "ቶ": 82,
+  "ቷ": 168,
+  "ቸ": 108,
+  "ቹ": 137,
+  "ቺ": 203,
+  "ቻ": 224,
+  "ቼ": 183,
+  "ች": 225,
+  "ቾ": 210,
+  "ቿ": 51,
+  "ኀ": 214,
+  "ኃ": 185,
+  "ኅ": 11,
+  "ኋ": 130,
+  "ነ": 177,
+  "ኑ": 204,
+  "ኒ": 30,
+  "ና": 49,
+  "ኔ": 45,
+  "ን": 14,
+  "ኖ": 112,
+  "ኗ": 145,
+  "ኘ": 70,
+  "ኙ": 174,
+  "ኛ": 64,
+  "ኝ": 59,
+  "ኞ": 38,
+  "ኟ": 180,
+  "አ": 19,
+  "ኡ": 53,
+  "ኢ": 29,
+  "ኤ": 213,
+  "እ": 101,
+  "ኦ": 153,
+  "ከ": 221,
+  "ኩ": 91,
+  "ኪ": 17,
+  "ካ": 12,
+  "ኬ": 106,
+  "ክ": 190,
+  "ኮ": 72,
+  "ኳ": 116,
+  "ኸ": 84,
+  "ኽ": 132,
+  "ወ": 85,
+  "ዊ": 122,
+  "ዋ": 50,
+  "ዌ": 201,
+  "ው": 0,
+  "ዎ": 121,
+  "ዐ": 73,
+  "ዑ": 155,
+  "ዒ": 178,
+  "ዓ": 26,
+  "ዕ": 69,
+  "ዖ": 152,
+  "ዘ": 118,
+  "ዙ": 146,
+  "ዚ": 76,
+  "ዛ": 206,
+  "ዜ": 28,
+  "ዝ": 127,
+  "ዞ": 115,
+  "ዟ": 104,
+  "ዢ": 83,
+  "ዣ": 219,
+  "ዤ": 125,
+  "ዥ": 140,
+  "ዦ": 99,
+  "የ": 57,
+  "ዩ": 62,
+  "ያ": 119,
+  "ዬ": 3,
+  "ይ": 211,
+  "ዮ": 33,
+  "ደ": 74,
+  "ዱ": 96,
+  "ዲ": 67,
+  "ዳ": 215,
+  "ዴ": 18,
+  "ድ": 103,
+  "ዶ": 10,
+  "ዷ": 154,
+  "ጀ": 117,
   "ጁ": 16,
-  "ጂ": 44,
-  "ጃ": 81,
-  "ጄ": 33,
-  "ጅ": 5,
-  "ጆ": 113,
-  "ገ": 212,
-  "ጉ": 56,
-  "ጊ": 160,
-  "ጋ": 69,
-  "��": 203,
-  "ግ": 35,
-  "ጎ": 199,
-  "ጓ": 40,
-  "ጠ": 162,
-  "ጡ": 49,
-  "ጢ": 112,
-  "ጣ": 99,
-  "ጤ": 158,
-  "ጥ": 37,
-  "ጦ": 225,
-  "ጧ": 189,
+  "ጂ": 89,
+  "ጃ": 227,
+  "ጄ": 61,
+  "ጅ": 86,
+  "ጆ": 181,
+  "ገ": 39,
+  "ጉ": 113,
+  "ጊ": 161,
+  "ጋ": 52,
+  "ጌ": 196,
+  "ግ": 66,
+  "ጎ": 217,
+  "ጓ": 166,
+  "ጠ": 41,
+  "ጡ": 143,
+  "ጢ": 205,
+  "ጣ": 209,
+  "ጤ": 4,
+  "ጥ": 160,
+  "ጦ": 93,
+  "ጧ": 193,
   "ጨ": 88,
-  "ጩ": 53,
-  "ጪ": 29,
-  "ጫ": 100,
-  "ጬ": 60,
-  "ጭ": 219,
-  "ጮ": 213,
-  "ጲ": 101,
-  "ጴ": 176,
-  "ጵ": 221,
-  "ጶ": 116,
-  "ጸ": 193,
-  "ጹ": 186,
-  "ጺ": 11,
-  "ጻ": 208,
-  "ጽ": 109,
-  "ጾ": 20,
-  "ጿ": 196,
-  "ፀ": 114,
-  "ፁ": 55,
-  "ፃ": 50,
-  "ፅ": 28,
-  "ፈ": 215,
-  "ፉ": 94,
-  "ፊ": 75,
-  "ፋ": 22,
-  "ፌ": 97,
-  "ፍ": 84,
-  "ፎ": 216,
-  "ፏ": 31,
-  "ፑ": 149,
-  "ፒ": 1,
-  "ፓ": 59,
-  "ፔ": 47,
-  "ፕ": 111,
-  "ፖ": 137,
-  "“": 124,
-  "”": 66
+  "ጩ": 54,
+  "ጪ": 56,
+  "ጫ": 199,
+  "ጬ": 162,
+  "ጭ": 6,
+  "ጮ": 32,
+  "ጲ": 21,
+  "ጴ": 128,
+  "ጵ": 149,
+  "ጶ": 68,
+  "ጸ": 47,
+  "ጹ": 31,
+  "ጺ": 179,
+  "ጻ": 98,
+  "ጽ": 37,
+  "ጾ": 42,
+  "ጿ": 189,
+  "ፀ": 65,
+  "ፁ": 100,
+  "ፃ": 216,
+  "ፅ": 212,
+  "ፈ": 48,
+  "ፉ": 15,
+  "ፊ": 135,
+  "ፋ": 111,
+  "ፌ": 71,
+  "ፍ": 147,
+  "ፎ": 109,
+  "ፏ": 159,
+  "ፑ": 124,
+  "ፒ": 25,
+  "ፓ": 133,
+  "ፔ": 131,
+  "ፕ": 171,
+  "ፖ": 92,
+  "“": 63,
+  "”": 20
 }