1 line
878 KiB
JSON
1 line
878 KiB
JSON
|
{"[PAD]": 0, "[CLS]": 1, "[SEP]": 2, "[UNK]": 3, "!": 328, "\"": 113, "#": 10431, "$": 1629, "%": 207, "&": 947, "'": 108, "(": 1640, ")": 43, "*": 3226, "+": 2744, ",": 6, "-": 12, ".": 4, "/": 73, "0": 288, "1": 134, "2": 176, "3": 246, "4": 306, "5": 245, "6": 401, "7": 406, "8": 398, "9": 466, ":": 35, ";": 131, "<": 41552, "=": 5214, ">": 15698, "?": 116, "@": 1039, "A": 250, "B": 387, "C": 347, "D": 495, "E": 717, "F": 597, "G": 534, "H": 725, "I": 100, "J": 863, "K": 530, "L": 574, "M": 448, "N": 487, "O": 673, "P": 510, "Q": 1864, "R": 500, "S": 104, "T": 565, "U": 791, "V": 846, "W": 771, "X": 1000, "Y": 975, "Z": 1301, "[": 10975, "\\": 37457, "]": 742, "^": 35227, "_": 1215, "`": 12905, "a": 102, "b": 428, "c": 438, "d": 417, "e": 242, "f": 506, "g": 571, "h": 298, "i": 118, "j": 267, "k": 330, "l": 462, "m": 119, "n": 282, "o": 139, "p": 642, "q": 1343, "r": 338, "s": 29, "t": 90, "u": 257, "v": 705, "w": 605, "x": 1178, "y": 219, "z": 329, "{": 45152, "|": 15483, "}": 24303, "~": 34437, "¡": 5543, "¢": 7258, "£": 2469, "¤": 10470, "¥": 8210, "¦": 18164, "§": 6248, "¨": 11423, "©": 15375, "ª": 10278, "«": 4958, "¬": 11582, "®": 2840, "¯": 10965, "°": 7487, "±": 15389, "²": 14292, "³": 15264, "´": 20024, "µ": 8906, "¶": 19002, "·": 18400, "¸": 18537, "¹": 9253, "º": 3070, "»": 2023, "¼": 4394, "½": 10809, "¾": 4726, "¿": 9470, "À": 50093, "Á": 50094, "Â": 4056, "Ã": 3849, "Ä": 649, "Å": 1277, "Æ": 8188, "Ç": 47040, "È": 6163, "É": 35423, "Ê": 38155, "Ë": 43621, "Ì": 44025, "Í": 48254, "Î": 41335, "Ï": 45704, "Ð": 25417, "Ñ": 22063, "Ò": 48561, "Ó": 49013, "Ô": 49077, "Õ": 42088, "Ö": 46030, "×": 42358, "Ø": 21593, "Ù": 26068, "Ú": 46164, "Û": 44148, "Ü": 49786, "Ý": 50011, "Þ": 49944, "ß": 49860, "à": 12798, "á": 1376, "â": 24987, "ã": 48064, "ä": 48753, "å": 42393, "æ": 37127, "ç": 36714, "è": 36484, "é": 41907, "ê": 46873, "ë": 45209, "ì": 43998, "í": 47649, "î": 49190, "ï": 43251, "ð": 49585, "ñ": 50095, "ò": 50096, "ó": 49078, "ô": 49820, "õ": 50097, "ö": 50098, "÷": 50099, "ø": 50100, "ù": 50101, "ú": 50102, "û": 50103, "ü": 50104, "ý": 50105, "þ": 50106, "ÿ": 50107, "Ā": 50108, "ā": 50109, "Ă": 50110, "ă": 50111, "Ą": 50112, "ą": 50113, "Ć": 50114, "ć": 50115, "Ĉ": 50116, "ĉ": 50117, "Ċ": 50118, "ċ": 50119, "Č": 50120, "č": 50121, "Ď": 50122, "ď": 50123, "Đ": 50124, "đ": 50125, "Ē": 50126, "ē": 50127, "Ĕ": 50128, "ĕ": 50129, "Ė": 50130, "ė": 50131, "Ę": 50132, "ę": 50133, "Ě": 50134, "ě": 50135, "Ĝ": 50136, "ĝ": 50137, "Ğ": 50138, "ğ": 50139, "Ġ": 1437, "ġ": 49802, "Ģ": 7471, "ģ": 10172, "Ĥ": 9264, "ĥ": 862, "Ħ": 11936, "ħ": 5782, "Ĩ": 27819, "ĩ": 6382, "Ī": 23133, "ī": 23171, "Ĭ": 27969, "ĭ": 13859, "Į": 14285, "į": 8384, "İ": 12736, "ı": 9357, "IJ": 16948, "ij": 3602, "Ĵ": 10659, "ĵ": 9085, "Ķ": 10674, "ķ": 15722, "ĸ": 25448, "Ĺ": 6800, "ĺ": 711, "Ļ": 27, "ļ": 15113, "Ľ": 3726, "ľ": 48, "Ŀ": 46, "ŀ": 17772, "Ł": 4333, "ł": 21402, "Ń": 12410, "Ġt": 326, "Ġa": 10, "he": 700, "in": 179, "re": 241, "on": 261, "Ġthe": 5, "er": 254, "Ġs": 579, "at": 415, "Ġw": 885, "Ġo": 1021, "en": 225, "Ġc": 740, "it": 405, "is": 354, "an": 260, "or": 368, "es": 293, "Ġb": 741, "ed": 196, "Ġf": 856, "ing": 154, "Ġp": 181, "ou": 1438, "Ġan": 41, "al": 337, "ar": 271, "Ġto": 7, "Ġm": 475, "Ġof": 9, "Ġin": 11, "Ġd": 385, "Ġh": 1368, "Ġand": 8, "ic": 636, "as": 281, "le": 459, "Ġth": 3553, "ion": 1499, "om": 1075, "ll": 890, "ent": 1342, "Ġn": 295, "Ġl": 784, "st": 620, "Ġre": 769, "ve": 548, "Ġe": 364, "ro": 1001, "ly": 352, "Ġbe": 28, "Ġg": 821, "ĠT": 255, "ct": 3894, "ĠS": 208, "id": 808, "ot": 1242, "ĠI": 38, "ut": 1182, "et": 594, "ĠA": 83, "Ġis": 16, "Ġon": 15, "im": 757, "am": 424, "ow": 1722, "ay": 857, "ad": 625, "se": 1090, "Ġthat": 14, "ĠC": 230, "ig": 1023, "Ġfor": 13, "ac": 1043, "Ġy": 1423, "ver": 2802, "ur": 710, "Ġu": 1717, "ld": 4779, "Ġst": 1690, "ĠM": 256, "'s": 18, "Ġhe": 37, "Ġit": 24, "ation": 1258, "ith": 3
|