Upload tokenizer

added_tokens.json

@@ -1,4 +1,4 @@
 {
-  "</s>": 97,
-  "<s>": 96
+  "</s>": 95,
+  "<s>": 94
 }

tokenizer_config.json

@@ -1,6 +1,6 @@
 {
   "added_tokens_decoder": {
-    "94": {
+    "92": {
       "content": "[UNK]",
       "lstrip": true,
       "normalized": false,
@@ -8,7 +8,7 @@
       "single_word": false,
       "special": false
     },
-    "95": {
+    "93": {
       "content": "[PAD]",
       "lstrip": true,
       "normalized": false,
@@ -16,7 +16,7 @@
       "single_word": false,
       "special": false
     },
-    "96": {
+    "94": {
       "content": "<s>",
       "lstrip": false,
       "normalized": true,
@@ -24,7 +24,7 @@
       "single_word": false,
       "special": true
     },
-    "97": {
+    "95": {
       "content": "</s>",
       "lstrip": false,
       "normalized": true,

vocab.json

@@ -1,6 +1,6 @@
 {
-  "[PAD]": 95,
-  "[UNK]": 94,
+  "[PAD]": 93,
+  "[UNK]": 92,
   "a": 1,
   "b": 2,
   "c": 3,
@@ -30,69 +30,67 @@
   "|": 0,
   "à": 27,
   "á": 28,
-  "â": 29,
-  "ã": 30,
-  "è": 31,
-  "é": 32,
-  "ê": 33,
-  "ì": 34,
-  "í": 35,
-  "ò": 36,
-  "ó": 37,
-  "ô": 38,
-  "õ": 39,
-  "ù": 40,
-  "ú": 41,
-  "ý": 42,
-  "ă": 43,
-  "đ": 44,
-  "ĩ": 45,
-  "ũ": 46,
-  "ơ": 47,
-  "ư": 48,
-  "ạ": 49,
-  "ả": 50,
-  "ấ": 51,
-  "ầ": 52,
-  "ẩ": 53,
-  "ẫ": 54,
-  "ậ": 55,
-  "ắ": 56,
-  "ằ": 57,
-  "ẳ": 58,
-  "ẵ": 59,
-  "ặ": 60,
-  "ẹ": 61,
-  "ẻ": 62,
-  "ẽ": 63,
-  "ế": 64,
-  "ề": 65,
-  "ể": 66,
-  "ễ": 67,
-  "ệ": 68,
-  "ỉ": 69,
-  "ị": 70,
-  "ọ": 71,
-  "ỏ": 72,
-  "ố": 73,
-  "ồ": 74,
-  "ổ": 75,
-  "ỗ": 76,
-  "ộ": 77,
-  "ớ": 78,
-  "ờ": 79,
-  "ở": 80,
-  "ỡ": 81,
-  "ợ": 82,
-  "ụ": 83,
-  "ủ": 84,
-  "ứ": 85,
-  "ừ": 86,
-  "ử": 87,
-  "ữ": 88,
-  "ự": 89,
-  "ỳ": 90,
-  "ỵ": 91,
-  "ỷ": 92,
-  "ỹ": 93
+  "ã": 29,
+  "è": 30,
+  "é": 31,
+  "ê": 32,
+  "ì": 33,
+  "í": 34,
+  "ò": 35,
+  "ó": 36,
+  "õ": 37,
+  "ù": 38,
+  "ú": 39,
+  "ý": 40,
+  "ă": 41,
+  "đ": 42,
+  "ĩ": 43,
+  "ũ": 44,
+  "ơ": 45,
+  "ư": 46,
+  "ạ": 47,
+  "ả": 48,
+  "ấ": 49,
+  "ầ": 50,
+  "ẩ": 51,
+  "ẫ": 52,
+  "ậ": 53,
+  "ắ": 54,
+  "ằ": 55,
+  "ẳ": 56,
+  "ẵ": 57,
+  "ặ": 58,
+  "ẹ": 59,
+  "ẻ": 60,
+  "ẽ": 61,
+  "ế": 62,
+  "ề": 63,
+  "ể": 64,
+  "ễ": 65,
+  "ệ": 66,
+  "ỉ": 67,
+  "ị": 68,
+  "ọ": 69,
+  "ỏ": 70,
+  "ố": 71,
+  "ồ": 72,
+  "ổ": 73,
+  "ỗ": 74,
+  "ộ": 75,
+  "ớ": 76,
+  "ờ": 77,
+  "ở": 78,
+  "ỡ": 79,
+  "ợ": 80,
+  "ụ": 81,
+  "ủ": 82,
+  "ứ": 83,
+  "ừ": 84,
+  "ử": 85,
+  "ữ": 86,
+  "ự": 87,
+  "ỳ": 88,
+  "ỵ": 89,
+  "ỷ": 90,
+  "ỹ": 91
 }