Upload processor

vocab.json

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