Upload processor

vocab.json

@@ -1 +1,112 @@
-{"%": 0, "\u00f2": 1, "\u1ed1": 2, "h": 4, "d": 5, "\u1eef": 6, "\u00e8": 7, "\u1ee1": 8, "b": 9, "7": 10, "\u00e2": 11, "\u1ed7": 12, "\u1eaf": 13, "\u00e3": 14, "4": 15, "\u0111": 16, "\u1ead": 17, "\u1ea9": 18, "\u1eab": 19, "u": 20, "\u1edd": 21, "l": 22, "\u0169": 23, "\u1ecf": 24, "\u1ef1": 25, "\u00f3": 26, "\u1ec1": 27, "c": 28, "\u1ef3": 29, "\u1ef7": 30, "\u0129": 31, "\u1ec5": 32, "a": 33, "\u00f9": 34, "\u1ebf": 35, "\u1ee7": 36, "m": 37, "\u1eb5": 38, "\u1eb7": 39, "6": 40, "\u00e9": 41, "n": 42, "\u1eb9": 43, "s": 44, "\u1ea5": 45, "\u1eed": 46, "t": 47, "k": 48, "o": 49, "\u1ebd": 50, "\u0300": 51, "0": 52, "\u00f4": 53, "\u0103": 54, "\u1ec3": 55, "9": 56, "\u1ee3": 57, "r": 58, "w": 59, "\u00ea": 60, "\u00fd": 61, "\u1ed9": 62, "f": 63, "\u0323": 64, "3": 65, "e": 66, "\u00ed": 67, "p": 68, "\u1ed5": 69, "\u1ea1": 70, "\u1edb": 71, "8": 72, "2": 73, "\u00ec": 74, "\u1ec7": 75, "\u00fa": 76, "q": 77, "\u00f5": 78, "i": 79, "\u1ec9": 80, "\u01b0": 81, "g": 82, "\u1ecb": 83, "\u1ea7": 84, "\u1ed3": 85, "\u1ee5": 86, "\u1ebb": 87, "\u1ef5": 88, "\u1eeb": 89, "\u01a1": 90, "y": 91, "\u1ecd": 92, "\u00e0": 93, "\u1eb3": 94, "\u0443": 95, "\u1edf": 96, "5": 97, "\u1ee9": 98, "z": 99, "\u1ea3": 100, "\u1eb1": 101, "v": 102, "\u0301": 103, "x": 104, "\u1ef9": 105, "1": 106, "\u00e1": 107, "|": 3, "[UNK]": 108, "[PAD]": 109}
+{
+  "%": 0,
+  "0": 52,
+  "1": 106,
+  "2": 73,
+  "3": 65,
+  "4": 15,
+  "5": 97,
+  "6": 40,
+  "7": 10,
+  "8": 72,
+  "9": 56,
+  "[PAD]": 109,
+  "[UNK]": 108,
+  "a": 33,
+  "b": 9,
+  "c": 28,
+  "d": 5,
+  "e": 66,
+  "f": 63,
+  "g": 82,
+  "h": 4,
+  "i": 79,
+  "k": 48,
+  "l": 22,
+  "m": 37,
+  "n": 42,
+  "o": 49,
+  "p": 68,
+  "q": 77,
+  "r": 58,
+  "s": 44,
+  "t": 47,
+  "u": 20,
+  "v": 102,
+  "w": 59,
+  "x": 104,
+  "y": 91,
+  "z": 99,
+  "|": 3,
+  "à": 93,
+  "á": 107,
+  "â": 11,
+  "ã": 14,
+  "è": 7,
+  "é": 41,
+  "ê": 60,
+  "ì": 74,
+  "í": 67,
+  "ò": 1,
+  "ó": 26,
+  "ô": 53,
+  "õ": 78,
+  "ù": 34,
+  "ú": 76,
+  "ý": 61,
+  "ă": 54,
+  "đ": 16,
+  "ĩ": 31,
+  "ũ": 23,
+  "ơ": 90,
+  "ư": 81,
+  "̀": 51,
+  "́": 103,
+  "̣": 64,
+  "у": 95,
+  "ạ": 70,
+  "ả": 100,
+  "ấ": 45,
+  "ầ": 84,
+  "ẩ": 18,
+  "ẫ": 19,
+  "ậ": 17,
+  "ắ": 13,
+  "ằ": 101,
+  "ẳ": 94,
+  "ẵ": 38,
+  "ặ": 39,
+  "ẹ": 43,
+  "ẻ": 87,
+  "ẽ": 50,
+  "ế": 35,
+  "ề": 27,
+  "ể": 55,
+  "ễ": 32,
+  "ệ": 75,
+  "ỉ": 80,
+  "ị": 83,
+  "ọ": 92,
+  "ỏ": 24,
+  "ố": 2,
+  "ồ": 85,
+  "ổ": 69,
+  "ỗ": 12,
+  "ộ": 62,
+  "ớ": 71,
+  "ờ": 21,
+  "ở": 96,
+  "ỡ": 8,
+  "ợ": 57,
+  "ụ": 86,
+  "ủ": 36,
+  "ứ": 98,
+  "ừ": 89,
+  "ử": 46,
+  "ữ": 6,
+  "ự": 25,
+  "ỳ": 29,
+  "ỵ": 88,
+  "ỷ": 30,
+  "ỹ": 105
+}