Upload processor

vocab.json

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