Upload processor

vocab.json

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