Upload tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 135,
-  "<s>": 134
+  "</s>": 115,
+  "<s>": 114
 }

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "132": {
+    "112": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "133": {
+    "113": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "134": {
+    "114": {
       "content": "<s>",
       "lstrip": false,
       "normalized": false,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "135": {
+    "115": {
       "content": "</s>",
       "lstrip": false,
       "normalized": false,

vocab.json

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