Upload tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 304,
-  "<s>": 303
+  "</s>": 289,
+  "<s>": 288
 }

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "301": {
+    "286": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "302": {
+    "287": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "303": {
+    "288": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "304": {
+    "289": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,

vocab.json

@@ -14,8 +14,8 @@
   "8": 13,
   "9": 14,
   "[": 15,
-  "[PAD]": 302,
-  "[UNK]": 301,
+  "[PAD]": 287,
+  "[UNK]": 286,
   "]": 16,
   "a": 17,
   "b": 18,
@@ -28,104 +28,104 @@
   "i": 25,
   "j": 26,
   "k": 27,
-  "m": 28,
-  "n": 29,
-  "o": 30,
-  "p": 31,
-  "r": 32,
-  "s": 33,
-  "t": 34,
-  "u": 35,
-  "v": 36,
-  "x": 37,
-  "y": 38,
+  "l": 28,
+  "m": 29,
+  "n": 30,
+  "o": 31,
+  "p": 32,
+  "q": 33,
+  "r": 34,
+  "s": 35,
+  "t": 36,
+  "u": 37,
+  "v": 38,
+  "w": 39,
+  "x": 40,
+  "y": 41,
+  "z": 42,
   "|": 0,
-  "~": 39,
-  "£": 40,
-  "¥": 41,
-  "°": 42,
-  "²": 43,
-  "½": 44,
-  "¾": 45,
-  "ã": 46,
-  "õ": 47,
-  "ሀ": 48,
-  "ሁ": 49,
-  "ሂ": 50,
-  "ሃ": 51,
-  "ሄ": 52,
-  "ህ": 53,
-  "ሆ": 54,
-  "ለ": 55,
-  "ሉ": 56,
-  "ሊ": 57,
-  "ላ": 58,
-  "ሌ": 59,
-  "ል": 60,
-  "ሎ": 61,
-  "ሏ": 62,
-  "ሐ": 63,
-  "ሑ": 64,
-  "ሓ": 65,
-  "ሔ": 66,
-  "ሕ": 67,
-  "መ": 68,
-  "ሙ": 69,
-  "ሚ": 70,
-  "ማ": 71,
-  "ሜ": 72,
-  "ም": 73,
-  "ሞ": 74,
-  "ሟ": 75,
-  "ሠ": 76,
-  "ሡ": 77,
-  "ሣ": 78,
-  "ሥ": 79,
-  "ሦ": 80,
-  "ረ": 81,
-  "ሩ": 82,
-  "ሪ": 83,
-  "ራ": 84,
-  "ሬ": 85,
-  "ር": 86,
-  "ሮ": 87,
-  "ሯ": 88,
-  "ሰ": 89,
-  "ሱ": 90,
-  "ሲ": 91,
-  "ሳ": 92,
-  "ሴ": 93,
-  "ስ": 94,
-  "ሶ": 95,
-  "ሷ": 96,
-  "ሸ": 97,
-  "ሹ": 98,
-  "ሺ": 99,
-  "ሻ": 100,
-  "ሼ": 101,
-  "ሽ": 102,
-  "ሾ": 103,
-  "ሿ": 104,
-  "ቀ": 105,
-  "ቁ": 106,
-  "ቂ": 107,
-  "ቃ": 108,
-  "ቄ": 109,
-  "ቅ": 110,
-  "ቆ": 111,
-  "ቋ": 112,
-  "በ": 113,
-  "ቡ": 114,
-  "ቢ": 115,
-  "ባ": 116,
-  "ቤ": 117,
-  "ብ": 118,
-  "ቦ": 119,
-  "ቧ": 120,
-  "ቨ": 121,
-  "ቪ": 122,
-  "ቫ": 123,
-  "ቬ": 124,
+  "~": 43,
+  "£": 44,
+  "¥": 45,
+  "°": 46,
+  "²": 47,
+  "½": 48,
+  "¾": 49,
+  "ã": 50,
+  "õ": 51,
+  "ሀ": 52,
+  "ሁ": 53,
+  "ሂ": 54,
+  "ሃ": 55,
+  "ሄ": 56,
+  "ህ": 57,
+  "ሆ": 58,
+  "ለ": 59,
+  "ሉ": 60,
+  "ሊ": 61,
+  "ላ": 62,
+  "ሌ": 63,
+  "ል": 64,
+  "ሎ": 65,
+  "ሏ": 66,
+  "ሐ": 67,
+  "ሓ": 68,
+  "ሔ": 69,
+  "ሕ": 70,
+  "መ": 71,
+  "ሙ": 72,
+  "ሚ": 73,
+  "ማ": 74,
+  "ሜ": 75,
+  "ም": 76,
+  "ሞ": 77,
+  "ሟ": 78,
+  "ሠ": 79,
+  "ሣ": 80,
+  "ሥ": 81,
+  "ሦ": 82,
+  "ረ": 83,
+  "ሩ": 84,
+  "ሪ": 85,
+  "ራ": 86,
+  "ሬ": 87,
+  "ር": 88,
+  "ሮ": 89,
+  "ሯ": 90,
+  "ሰ": 91,
+  "ሱ": 92,
+  "ሲ": 93,
+  "ሳ": 94,
+  "ሴ": 95,
+  "ስ": 96,
+  "ሶ": 97,
+  "ሷ": 98,
+  "ሸ": 99,
+  "ሹ": 100,
+  "ሺ": 101,
+  "ሻ": 102,
+  "ሼ": 103,
+  "ሽ": 104,
+  "ሾ": 105,
+  "ቀ": 106,
+  "ቁ": 107,
+  "ቂ": 108,
+  "ቃ": 109,
+  "ቄ": 110,
+  "ቅ": 111,
+  "ቆ": 112,
+  "ቋ": 113,
+  "በ": 114,
+  "ቡ": 115,
+  "ቢ": 116,
+  "ባ": 117,
+  "ቤ": 118,
+  "ብ": 119,
+  "ቦ": 120,
+  "ቧ": 121,
+  "ቨ": 122,
+  "ቪ": 123,
+  "ቫ": 124,
   "ቭ": 125,
   "ቮ": 126,
   "ተ": 127,
@@ -160,146 +160,131 @@
   "ኙ": 156,
   "ኚ": 157,
   "ኛ": 158,
-  "ኜ": 159,
-  "ኝ": 160,
-  "ኞ": 161,
-  "ኟ": 162,
-  "አ": 163,
-  "ኡ": 164,
-  "ኢ": 165,
-  "ኣ": 166,
-  "ኤ": 167,
-  "እ": 168,
-  "ኦ": 169,
-  "ከ": 170,
-  "ኩ": 171,
-  "ኪ": 172,
-  "ካ": 173,
-  "ኬ": 174,
-  "ክ": 175,
-  "ኮ": 176,
-  "ኰ": 177,
-  "ኳ": 178,
-  "ኸ": 179,
-  "ኺ": 180,
-  "ኽ": 181,
-  "ወ": 182,
-  "ዉ": 183,
-  "ዊ": 184,
-  "ዋ": 185,
-  "ዌ": 186,
-  "ው": 187,
-  "ዎ": 188,
-  "ዐ": 189,
-  "ዑ": 190,
-  "ዒ": 191,
-  "ዓ": 192,
-  "ዕ": 193,
-  "ዖ": 194,
-  "ዘ": 195,
-  "ዙ": 196,
-  "ዚ": 197,
-  "ዛ": 198,
-  "ዜ": 199,
-  "ዝ": 200,
-  "ዞ": 201,
-  "ዟ": 202,
-  "ዠ": 203,
-  "ዡ": 204,
-  "ዢ": 205,
-  "ዣ": 206,
-  "ዤ": 207,
-  "ዥ": 208,
-  "ዦ": 209,
-  "የ": 210,
-  "ዩ": 211,
-  "ዪ": 212,
-  "ያ": 213,
-  "ዬ": 214,
-  "ይ": 215,
-  "ዮ": 216,
-  "ደ": 217,
-  "ዱ": 218,
-  "ዲ": 219,
-  "ዳ": 220,
-  "ዴ": 221,
-  "ድ": 222,
-  "ዶ": 223,
-  "ዷ": 224,
-  "ጀ": 225,
-  "ጁ": 226,
-  "ጂ": 227,
-  "ጃ": 228,
-  "ጄ": 229,
-  "ጅ": 230,
-  "ጆ": 231,
-  "ጇ": 232,
-  "ገ": 233,
-  "ጉ": 234,
-  "ጊ": 235,
-  "ጋ": 236,
-  "ጌ": 237,
-  "ግ": 238,
-  "ጎ": 239,
-  "ጐ": 240,
-  "ጓ": 241,
-  "ጠ": 242,
-  "ጡ": 243,
-  "ጢ": 244,
-  "ጣ": 245,
-  "ጤ": 246,
-  "ጥ": 247,
-  "ጦ": 248,
-  "ጧ": 249,
-  "ጨ": 250,
-  "ጩ": 251,
-  "ጪ": 252,
-  "ጫ": 253,
-  "ጬ": 254,
-  "ጭ": 255,
-  "ጮ": 256,
-  "ጲ": 257,
-  "ጳ": 258,
-  "ጴ": 259,
-  "ጵ": 260,
-  "ጶ": 261,
-  "ጸ": 262,
-  "ጹ": 263,
-  "ጺ": 264,
-  "ጻ": 265,
-  "ጽ": 266,
-  "ጾ": 267,
-  "ጿ": 268,
-  "ፀ": 269,
-  "ፁ": 270,
-  "ፃ": 271,
-  "ፅ": 272,
-  "ፆ": 273,
-  "ፈ": 274,
-  "ፉ": 275,
-  "ፊ": 276,
-  "ፋ": 277,
-  "ፌ": 278,
-  "ፍ": 279,
-  "ፎ": 280,
-  "ፏ": 281,
-  "ፐ": 282,
-  "ፑ": 283,
-  "ፒ": 284,
-  "ፓ": 285,
-  "ፔ": 286,
-  "ፕ": 287,
-  "ፖ": 288,
-  "ፗ": 289,
-  "፡": 290,
-  "።": 291,
-  "፣": 292,
-  "፤": 293,
-  "፥": 294,
-  "፦": 295,
-  "–": 296,
-  "—": 297,
-  "’": 298,
-  "‹": 299,
-  "›": 300
+  "ኝ": 159,
+  "ኞ": 160,
+  "አ": 161,
+  "ኡ": 162,
+  "ኢ": 163,
+  "ኣ": 164,
+  "ኤ": 165,
+  "እ": 166,
+  "ኦ": 167,
+  "ከ": 168,
+  "ኩ": 169,
+  "ኪ": 170,
+  "ካ": 171,
+  "ኬ": 172,
+  "ክ": 173,
+  "ኮ": 174,
+  "ኳ": 175,
+  "ኸ": 176,
+  "ወ": 177,
+  "ዉ": 178,
+  "ዊ": 179,
+  "ዋ": 180,
+  "ዌ": 181,
+  "ው": 182,
+  "ዎ": 183,
+  "ዐ": 184,
+  "ዑ": 185,
+  "ዓ": 186,
+  "ዕ": 187,
+  "ዖ": 188,
+  "ዘ": 189,
+  "ዙ": 190,
+  "ዚ": 191,
+  "ዛ": 192,
+  "ዜ": 193,
+  "ዝ": 194,
+  "ዞ": 195,
+  "ዟ": 196,
+  "ዠ": 197,
+  "ዢ": 198,
+  "ዣ": 199,
+  "ዥ": 200,
+  "ዦ": 201,
+  "የ": 202,
+  "ዩ": 203,
+  "ዪ": 204,
+  "ያ": 205,
+  "ዬ": 206,
+  "ይ": 207,
+  "ዮ": 208,
+  "ደ": 209,
+  "ዱ": 210,
+  "ዲ": 211,
+  "ዳ": 212,
+  "ዴ": 213,
+  "ድ": 214,
+  "ዶ": 215,
+  "ዷ": 216,
+  "ጀ": 217,
+  "ጁ": 218,
+  "ጂ": 219,
+  "ጃ": 220,
+  "ጄ": 221,
+  "ጅ": 222,
+  "ጆ": 223,
+  "ጇ": 224,
+  "ገ": 225,
+  "ጉ": 226,
+  "ጊ": 227,
+  "ጋ": 228,
+  "ጌ": 229,
+  "ግ": 230,
+  "ጎ": 231,
+  "ጓ": 232,
+  "ጠ": 233,
+  "ጡ": 234,
+  "ጢ": 235,
+  "ጣ": 236,
+  "ጤ": 237,
+  "ጥ": 238,
+  "ጦ": 239,
+  "ጧ": 240,
+  "ጨ": 241,
+  "ጩ": 242,
+  "ጪ": 243,
+  "ጫ": 244,
+  "ጭ": 245,
+  "ጮ": 246,
+  "ጳ": 247,
+  "ጴ": 248,
+  "ጵ": 249,
+  "ጸ": 250,
+  "ጹ": 251,
+  "ጻ": 252,
+  "ጽ": 253,
+  "ጾ": 254,
+  "ጿ": 255,
+  "ፀ": 256,
+  "ፁ": 257,
+  "ፃ": 258,
+  "ፅ": 259,
+  "ፈ": 260,
+  "ፉ": 261,
+  "ፊ": 262,
+  "ፋ": 263,
+  "ፌ": 264,
+  "ፍ": 265,
+  "ፎ": 266,
+  "ፏ": 267,
+  "ፐ": 268,
+  "ፑ": 269,
+  "ፒ": 270,
+  "ፓ": 271,
+  "ፔ": 272,
+  "ፕ": 273,
+  "ፖ": 274,
+  "፡": 275,
+  "።": 276,
+  "፣": 277,
+  "፤": 278,
+  "፥": 279,
+  "፦": 280,
+  "–": 281,
+  "—": 282,
+  "’": 283,
+  "‹": 284,
+  "›": 285
 }