Upload processor

vocab.json

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