Upload tokenizer

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "<unk>": 95
+}

special_tokens_map.json

@@ -0,0 +1,4 @@
+{
+  "pad_token": "ụ",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,32 @@
+{
+  "add_blank": true,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "ụ",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "95": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "extra_special_tokens": {},
+  "is_uroman": false,
+  "language": "vie",
+  "model_max_length": 1000000000000000019884624838656,
+  "normalize": true,
+  "pad_token": "ụ",
+  "phonemize": false,
+  "tokenizer_class": "VitsTokenizer",
+  "unk_token": "<unk>",
+  "verbose": false
+}

vocab.json

@@ -0,0 +1,97 @@
+{
+  " ": 84,
+  "'": 44,
+  "-": 94,
+  "2": 52,
+  "_": 17,
+  "a": 29,
+  "b": 88,
+  "c": 13,
+  "d": 63,
+  "e": 54,
+  "g": 21,
+  "h": 85,
+  "i": 30,
+  "k": 79,
+  "l": 82,
+  "m": 68,
+  "n": 90,
+  "o": 31,
+  "p": 78,
+  "q": 47,
+  "r": 92,
+  "s": 2,
+  "t": 80,
+  "u": 8,
+  "v": 14,
+  "x": 1,
+  "y": 75,
+  "à": 35,
+  "á": 77,
+  "â": 12,
+  "ã": 51,
+  "è": 3,
+  "é": 58,
+  "ê": 91,
+  "ì": 4,
+  "í": 74,
+  "ò": 45,
+  "ó": 56,
+  "ô": 28,
+  "õ": 25,
+  "ù": 38,
+  "ú": 76,
+  "ý": 37,
+  "ă": 89,
+  "đ": 55,
+  "ĩ": 23,
+  "ũ": 70,
+  "ơ": 7,
+  "ư": 9,
+  "ạ": 22,
+  "ả": 24,
+  "ấ": 81,
+  "ầ": 57,
+  "ẩ": 49,
+  "ẫ": 67,
+  "ậ": 87,
+  "ắ": 65,
+  "ằ": 10,
+  "ẳ": 27,
+  "ẵ": 42,
+  "ặ": 5,
+  "ẹ": 72,
+  "ẻ": 20,
+  "ẽ": 66,
+  "ế": 60,
+  "ề": 40,
+  "ể": 69,
+  "ễ": 41,
+  "ệ": 15,
+  "ỉ": 71,
+  "ị": 53,
+  "ọ": 48,
+  "ỏ": 43,
+  "ố": 46,
+  "ồ": 16,
+  "ổ": 34,
+  "ỗ": 73,
+  "ộ": 19,
+  "ớ": 59,
+  "ờ": 36,
+  "ở": 83,
+  "ỡ": 26,
+  "ợ": 93,
+  "ụ": 0,
+  "ủ": 61,
+  "ứ": 6,
+  "ừ": 32,
+  "ử": 62,
+  "ữ": 64,
+  "ự": 50,
+  "ỳ": 11,
+  "ỵ": 18,
+  "ỷ": 86,
+  "ỹ": 33,
+  "–": 39
+}