Upload processor

vocab.json

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