Upload tokenizer

added_tokens.json

@@ -1,5 +1,5 @@
 {
-  "</s>": 106,
-  "<s>": 105,
-  "[UNK]": 107
+  "</s>": 136,
+  "<s>": 135,
+  "[UNK]": 137
 }

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "103": {
+    "133": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "104": {
+    "134": {
       "content": "[UNK",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "105": {
+    "135": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "106": {
+    "136": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,
@@ -32,7 +32,7 @@
       "single_word": false,
       "special": true
     },
-    "107": {
+    "137": {
       "content": "[UNK]",
       "lstrip": false,
       "normalized": false,

vocab.json

@@ -1,8 +1,14 @@
 {
-  " ": 99,
-  ".": 88,
-  "[PAD]": 103,
-  "[UNK": 104,
+  " ": 110,
+  "(": 99,
+  ")": 95,
+  ".": 93,
+  "2": 107,
+  "4": 109,
+  "5": 104,
+  "6": 105,
+  "[PAD]": 133,
+  "[UNK": 134,
   "a": 84,
   "b": 1,
   "c": 6,
@@ -25,43 +31,43 @@
   "t": 2,
   "u": 65,
   "v": 30,
-  "w": 93,
+  "w": 100,
   "x": 41,
   "y": 61,
   "z": 34,
-  "|": 102,
-  "ä": 92,
+  "|": 132,
+  "ä": 98,
   "æ": 82,
   "ç": 39,
   "ð": 32,
   "ø": 70,
   "ħ": 45,
-  "ĩ": 101,
+  "ĩ": 131,
   "ŋ": 19,
   "œ": 77,
-  "ũ": 100,
+  "ũ": 130,
   "ɐ": 83,
   "ɑ": 86,
   "ɒ": 87,
   "ɔ": 81,
-  "ɕ": 89,
+  "ɕ": 94,
   "ɖ": 5,
-  "ɗ": 97,
+  "ɗ": 106,
   "ɘ": 71,
   "ə": 75,
-  "ɚ": 90,
+  "ɚ": 96,
   "ɛ": 76,
   "ɜ": 78,
   "ɞ": 79,
   "ɟ": 7,
-  "ɡ": 95,
+  "ɡ": 102,
   "ɢ": 11,
   "ɣ": 42,
   "ɤ": 73,
   "ɦ": 48,
   "ɨ": 62,
   "ɪ": 66,
-  "ɫ": 98,
+  "ɫ": 108,
   "ɬ": 49,
   "ɭ": 57,
   "ɮ": 50,
@@ -90,7 +96,7 @@
   "ʎ": 58,
   "ʏ": 67,
   "ʐ": 38,
-  "ʑ": 96,
+  "ʑ": 103,
   "ʒ": 36,
   "ʔ": 13,
   "ʕ": 46,
@@ -98,10 +104,34 @@
   "ʝ": 40,
   "ʟ": 59,
   "ʡ": 12,
-  "ː": 91,
+  "ʰ": 88,
+  "ʲ": 90,
+  "ʷ": 89,
+  "ː": 97,
+  "ˠ": 91,
+  "ˤ": 92,
+  "̃": 129,
+  "̈": 118,
+  "̊": 112,
+  "̘": 124,
+  "̙": 125,
+  "̜": 115,
+  "̝": 123,
+  "̞": 122,
+  "̟": 116,
+  "̠": 117,
+  "̤": 126,
+  "̥": 111,
+  "̩": 120,
+  "̬": 113,
+  "̯": 121,
+  "̰": 127,
+  "̹": 114,
+  "̼": 128,
+  "̽": 119,
   "β": 28,
   "θ": 31,
   "χ": 43,
-  "ᵝ": 94,
+  "ᵝ": 101,
   "ⱱ": 24
 }