add tokenizer

added_tokens.json

@@ -0,0 +1 @@
+{"<s>": 1434, "</s>": 1435}

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]", "additional_special_tokens": [{"content": "<s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, {"content": "</s>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}]}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "./", "tokenizer_class": "Wav2Vec2CTCTokenizer"}

vocab.json

@@ -0,0 +1 @@
+{"…": 1, "々": 2, "〇": 3, "「": 4, "」": 5, "あ": 6, "い": 7, "う": 8, "え": 9, "お": 10, "か": 11, "が": 12, "き": 13, "ぎ": 14, "く": 15, "ぐ": 16, "け": 17, "げ": 18, "こ": 19, "ご": 20, "さ": 21, "ざ": 22, "し": 23, "じ": 24, "す": 25, "ず": 26, "せ": 27, "ぜ": 28, "そ": 29, "ぞ": 30, "た": 31, "だ": 32, "ち": 33, "っ": 34, "つ": 35, "づ": 36, "て": 37, "で": 38, "と": 39, "ど": 40, "な": 41, "に": 42, "ぬ": 43, "ね": 44, "の": 45, "は": 46, "ば": 47, "ぱ": 48, "ひ": 49, "び": 50, "ぴ": 51, "ふ": 52, "ぶ": 53, "ぷ": 54, "へ": 55, "べ": 56, "ぺ": 57, "ほ": 58, "ぼ": 59, "ま": 60, "み": 61, "む": 62, "め": 63, "も": 64, "ゃ": 65, "や": 66, "ゅ": 67, "ゆ": 68, "ょ": 69, "よ": 70, "ら": 71, "り": 72, "る": 73, "れ": 74, "ろ": 75, "わ": 76, "を": 77, "ん": 78, "ァ": 79, "ア": 80, "ィ": 81, "イ": 82, "ウ": 83, "ェ": 84, "エ": 85, "ォ": 86, "オ": 87, "カ": 88, "ガ": 89, "キ": 90, "ギ": 91, "ク": 92, "グ": 93, "ケ": 94, "ゲ": 95, "コ": 96, "ゴ": 97, "サ": 98, "ザ": 99, "シ": 100, "ジ": 101, "ス": 102, "ズ": 103, "セ": 104, "ゼ": 105, "ソ": 106, "タ": 107, "ダ": 108, "チ": 109, "ッ": 110, "ツ": 111, "テ": 112, "デ": 113, "ト": 114, "ド": 115, "ナ": 116, "ニ": 117, "ヌ": 118, "ネ": 119, "ノ": 120, "ハ": 121, "バ": 122, "パ": 123, "ヒ": 124, "ビ": 125, "ピ": 126, "フ": 127, "ブ": 128, "プ": 129, "ヘ": 130, "ベ": 131, "ペ": 132, "ホ": 133, "ボ": 134, "ポ": 135, "マ": 136, "ミ": 137, "ム": 138, "メ": 139, "モ": 140, "ャ": 141, "ヤ": 142, "ュ": 143, "ユ": 144, "ョ": 145, "ヨ": 146, "ラ": 147, "リ": 148, "ル": 149, "レ": 150, "ロ": 151, "ワ": 152, "ン": 153, "ヴ": 154, "ヶ": 155, "ー": 156, "一": 157, "丁": 158, "七": 159, "万": 160, "丈": 161, "三": 162, "上": 163, "下": 164, "不": 165, "与": 166, "世": 167, "両": 168, "並": 169, "中": 170, "丸": 171, "主": 172, "久": 173, "乗": 174, "九": 175, "乱": 176, "乳": 177, "乾": 178, "了": 179, "予": 180, "争": 181, "事": 182, "二": 183, "互": 184, "五": 185, "井": 186, "交": 187, "京": 188, "人": 189, "今": 190, "介": 191, "仏": 192, "仕": 193, "他": 194, "付": 195, "代": 196, "以": 197, "件": 198, "任": 199, "企": 200, "伏": 201, "休": 202, "会": 203, "伝": 204, "伴": 205, "伸": 206, "似": 207, "位": 208, "低": 209, "住": 210, "体": 211, "何": 212, "余": 213, "作": 214, "併": 215, "使": 216, "供": 217, "依": 218, "価": 219, "侵": 220, "便": 221, "係": 222, "保": 223, "信": 224, "修": 225, "俵": 226, "俺": 227, "倉": 228, "個": 229, "倍": 230, "倒": 231, "候": 232, "借": 233, "倫": 234, "偉": 235, "健": 236, "側": 237, "偶": 238, "偽": 239, "傘": 240, "備": 241, "催": 242, "傷": 243, "働": 244, "像": 245, "僕": 246, "償": 247, "優": 248, "元": 249, "兄": 250, "充": 251, "先": 252, "光": 253, "免": 254, "児": 255, "入": 256, "全": 257, "八": 258, "公": 259, "六": 260, "共": 261, "具": 262, "兼": 263, "内": 264, "円": 265, "冊": 266, "再": 267, "冒": 268, "写": 269, "冠": 270, "冬": 271, "冷": 272, "凍": 273, "几": 274, "凡": 275, "処": 276, "出": 277, "分": 278, "切": 279, "刊": 280, "刑": 281, "列": 282, "初": 283, "別": 284, "利": 285, "制": 286, "券": 287, "刺": 288, "則": 289, "前": 290, "剤": 291, "剥": 292, "割": 293, "劇": 294, "力": 295, "功": 296, "加": 297, "劣": 298, "助": 299, "努": 300, "労": 301, "効": 302, "勇": 303, "勉": 304, "動": 305, "務": 306, "勝": 307, "勢": 308, "勤": 309, "包": 310, "化": 311, "北": 312, "匠": 313, "匹": 314, "区": 315, "医": 316, "十": 317, "千": 318, "午": 319, "半": 320, "卒": 321, "協": 322, "南": 323, "単": 324, "占": 325, "印": 326, "危": 327, "却": 328, "卵": 329, "厚": 330, "原": 331, "厳": 332, "去": 333, "参": 334, "及": 335, "友": 336, "反": 337, "収": 338, "叔": 339, "取": 340, "受": 341, "口": 342, "古": 343, "叩": 344, "可": 345, "台": 346, "史": 347, "右": 348, "号": 349, "司": 350, "各": 351, "合": 352, "吉": 353, "同": 354, "名": 355, "吐": 356, "向": 357, "君": 358, "否": 359, "含": 360, "吸": 361, "吹": 362, "呆": 363, "告": 364, "周": 365, "味": 366, "呼": 367, "命": 368, "和": 369, "咲": 370, "品": 371, "員": 372, "唯": 373, "商": 374, "問": 375, "善": 376, "喜": 377, "喫": 378, "喰": 379, "営": 380, "噂": 381, "器": 382, "四": 383, "回": 384, "因": 385, "団": 386, "困": 387, "囲": 388, "図": 389, "固": 390, "国": 391, "園": 392, "土": 393, "圧": 394, "在": 395, "地": 396, "坂": 397, "均": 398, "坊": 399, "型": 400, "垢": 401, "埋": 402, "城": 403, "域": 404, "堂": 405, "報": 406, "場": 407, "塀": 408, "塁": 409, "塩": 410, "塵": 411, "境": 412, "墓": 413, "増": 414, "壁": 415, "壊": 416, "士": 417, "壮": 418, "声": 419, "売": 420, "変": 421, "夏": 422, "夕": 423, "外": 424, "多": 425, "夜": 426, "夢": 427, "大": 428, "天": 429, "太": 430, "夫": 431, "央": 432, "失": 433, "奇": 434, "奈": 435, "奏": 436, "奥": 437, "女": 438, "好": 439, "妃": 440, "妙": 441, "妬": 442, "妹": 443, "妻": 444, "姉": 445, "始": 446, "姿": 447, "威": 448, "娘": 449, "婚": 450, "婦": 451, "嫌": 452, "嬉": 453, "子": 454, "字": 455, "存": 456, "季": 457, "学": 458, "孫": 459, "宅": 460, "宇": 461, "守": 462, "安": 463, "完": 464, "官": 465, "宙": 466, "定": 467, "宝": 468, "実": 469, "客": 470, "室": 471, "宮": 472, "害": 473, "宴": 474, "家": 475, "容": 476, "宿": 477, "寄": 478, "密": 479, "富": 480, "寒": 481, "寓": 482, "寝": 483, "寵": 484, "寸": 485, "寺": 486, "対": 487, "寿": 488, "封": 489, "専": 490, "射": 491, "将": 492, "尋": 493, "導": 494, "小": 495, "少": 496, "尺": 497, "局": 498, "屈": 499, "届": 500, "屋": 501, "属": 502, "層": 503, "山": 504, "岡": 505, "岩": 506, "岬": 507, "岸": 508, "島": 509, "崩": 510, "嶋": 511, "川": 512, "州": 513, "巡": 514, "工": 515, "左": 516, "巧": 517, "差": 518, "己": 519, "巻": 520, "市": 521, "布": 522, "希": 523, "帝": 524, "師": 525, "席": 526, "帯": 527, "帰": 528, "帳": 529, "常": 530, "帽": 531, "幅": 532, "幌": 533, "干": 534, "平": 535, "年": 536, "幸": 537, "幼": 538, "庁": 539, "広": 540, "床": 541, "底": 542, "店": 543, "府": 544, "度": 545, "座": 546, "庫": 547, "庭": 548, "康": 549, "廊": 550, "延": 551, "建": 552, "弁": 553, "弊": 554, "式": 555, "引": 556, "弟": 557, "弱": 558, "張": 559, "強": 560, "弾": 561, "当": 562, "形": 563, "彩": 564, "影": 565, "役": 566, "彼": 567, "往": 568, "待": 569, "後": 570, "徐": 571, "徒": 572, "従": 573, "得": 574, "御": 575, "復": 576, "微": 577, "徴": 578, "徹": 579, "心": 580, "必": 581, "忍": 582, "志": 583, "忘": 584, "忙": 585, "応": 586, "忠": 587, "念": 588, "怒": 589, "怖": 590, "思": 591, "怠": 592, "急": 593, "性": 594, "怪": 595, "恋": 596, "恐": 597, "息": 598, "恵": 599, "悔": 600, "患": 601, "悩": 602, "悪": 603, "悲": 604, "情": 605, "惑": 606, "想": 607, "意": 608, "愛": 609, "感": 610, "態": 611, "慌": 612, "慕": 613, "慢": 614, "慮": 615, "慰": 616, "憐": 617, "憩": 618, "憶": 619, "懐": 620, "懸": 621, "懺": 622, "成": 623, "我": 624, "戦": 625, "戸": 626, "戻": 627, "房": 628, "所": 629, "扇": 630, "扉": 631, "手": 632, "打": 633, "払": 634, "扱": 635, "扶": 636, "承": 637, "技": 638, "抑": 639, "抒": 640, "投": 641, "折": 642, "抜": 643, "押": 644, "拍": 645, "招": 646, "拝": 647, "拠": 648, "拡": 649, "拷": 650, "持": 651, "指": 652, "挙": 653, "振": 654, "捕": 655, "捨": 656, "掃": 657, "授": 658, "掌": 659, "排": 660, "掘": 661, "掛": 662, "採": 663, "探": 664, "接": 665, "描": 666, "揚": 667, "握": 668, "揮": 669, "援": 670, "揺": 671, "搬": 672, "携": 673, "撃": 674, "撮": 675, "擦": 676, "擬": 677, "支": 678, "改": 679, "攻": 680, "放": 681, "政": 682, "故": 683, "救": 684, "敗": 685, "教": 686, "散": 687, "数": 688, "整": 689, "敷": 690, "文": 691, "料": 692, "新": 693, "方": 694, "施": 695, "旅": 696, "旋": 697, "族": 698, "既": 699, "日": 700, "旧": 701, "早": 702, "昇": 703, "明": 704, "易": 705, "昔": 706, "星": 707, "映": 708, "春": 709, "昭": 710, "昼": 711, "時": 712, "晩": 713, "普": 714, "景": 715, "晴": 716, "暇": 717, "暑": 718, "暖": 719, "暗": 720, "暮": 721, "暴": 722, "曇": 723, "曜": 724, "曲": 725, "更": 726, "書": 727, "替": 728, "最": 729, "月": 730, "有": 731, "服": 732, "望": 733, "朝": 734, "期": 735, "木": 736, "末": 737, "本": 738, "札": 739, "机": 740, "材": 741, "村": 742, "束": 743, "条": 744, "来": 745, "杯": 746, "東": 747, "松": 748, "板": 749, "林": 750, "枚": 751, "果": 752, "枝": 753, "枠": 754, "柄": 755, "染": 756, "柔": 757, "柱": 758, "査": 759, "栄": 760, "栓": 761, "校": 762, "核": 763, "根": 764, "格": 765, "案": 766, "械": 767, "棋": 768, "棚": 769, "森": 770, "植": 771, "検": 772, "業": 773, "極": 774, "楽": 775, "概": 776, "構": 777, "様": 778, "標": 779, "模": 780, "横": 781, "樫": 782, "橋": 783, "機": 784, "欠": 785, "次": 786, "欧": 787, "欲": 788, "歌": 789, "歓": 790, "止": 791, "正": 792, "歩": 793, "歯": 794, "歳": 795, "歴": 796, "死": 797, "殊": 798, "残": 799, "殖": 800, "段": 801, "殻": 802, "殿": 803, "母": 804, "毎": 805, "毒": 806, "比": 807, "毛": 808, "氏": 809, "民": 810, "気": 811, "水": 812, "氷": 813, "永": 814, "氾": 815, "求": 816, "汗": 817, "汚": 818, "池": 819, "決": 820, "沈": 821, "河": 822, "油": 823, "治": 824, "沼": 825, "沿": 826, "況": 827, "泊": 828, "法": 829, "波": 830, "泣": 831, "泥": 832, "注": 833, "泳": 834, "洋": 835, "洒": 836, "洗": 837, "活": 838, "派": 839, "流": 840, "浄": 841, "浅": 842, "浜": 843, "浮": 844, "浴": 845, "海": 846, "浸": 847, "消": 848, "涙": 849, "涼": 850, "淋": 851, "淡": 852, "深": 853, "淵": 854, "混": 855, "済": 856, "減": 857, "渠": 858, "渡": 859, "温": 860, "港": 861, "湖": 862, "湯": 863, "満": 864, "源": 865, "準": 866, "溜": 867, "溢": 868, "滅": 869, "滑": 870, "滴": 871, "漁": 872, "漆": 873, "漏": 874, "演": 875, "漢": 876, "潜": 877, "潤": 878, "激": 879, "濃": 880, "濫": 881, "濯": 882, "瀬": 883, "火": 884, "灯": 885, "灰": 886, "炎": 887, "点": 888, "為": 889, "無": 890, "焦": 891, "然": 892, "焼": 893, "照": 894, "熊": 895, "熱": 896, "燃": 897, "燐": 898, "爆": 899, "父": 900, "片": 901, "版": 902, "牛": 903, "物": 904, "牲": 905, "特": 906, "牽": 907, "犠": 908, "犬": 909, "状": 910, "狩": 911, "独": 912, "狭": 913, "猟": 914, "猫": 915, "獅": 916, "獲": 917, "玄": 918, "率": 919, "玉": 920, "王": 921, "珍": 922, "現": 923, "球": 924, "理": 925, "琉": 926, "璧": 927, "瓶": 928, "甘": 929, "生": 930, "産": 931, "用": 932, "田": 933, "由": 934, "男": 935, "町": 936, "画": 937, "界": 938, "畑": 939, "留": 940, "略": 941, "番": 942, "異": 943, "畳": 944, "疲": 945, "病": 946, "症": 947, "痛": 948, "痢": 949, "痴": 950, "療": 951, "癖": 952, "発": 953, "登": 954, "白": 955, "百": 956, "的": 957, "皆": 958, "皮": 959, "皿": 960, "盛": 961, "監": 962, "盤": 963, "目": 964, "直": 965, "相": 966, "省": 967, "県": 968, "真": 969, "眠": 970, "眺": 971, "眼": 972, "着": 973, "睡": 974, "瞬": 975, "瞳": 976, "知": 977, "短": 978, "石": 979, "砂": 980, "研": 981, "砲": 982, "破": 983, "硬": 984, "碑": 985, "確": 986, "磨": 987, "示": 988, "社": 989, "祖": 990, "神": 991, "票": 992, "祭": 993, "福": 994, "禰": 995, "秀": 996, "私": 997, "秋": 998, "秘": 999, "称": 1000, "移": 1001, "程": 1002, "税": 1003, "種": 1004, "稿": 1005, "積": 1006, "穴": 1007, "究": 1008, "空": 1009, "突": 1010, "窒": 1011, "窓": 1012, "立": 1013, "章": 1014, "端": 1015, "競": 1016, "竹": 1017, "笑": 1018, "笛": 1019, "符": 1020, "筆": 1021, "筋": 1022, "筒": 1023, "答": 1024, "策": 1025, "算": 1026, "管": 1027, "箱": 1028, "節": 1029, "築": 1030, "米": 1031, "粉": 1032, "粧": 1033, "精": 1034, "糖": 1035, "糸": 1036, "系": 1037, "糾": 1038, "約": 1039, "紅": 1040, "紋": 1041, "純": 1042, "紙": 1043, "級": 1044, "紛": 1045, "素": 1046, "紫": 1047, "細": 1048, "終": 1049, "組": 1050, "経": 1051, "結": 1052, "絡": 1053, "給": 1054, "絵": 1055, "絶": 1056, "絹": 1057, "継": 1058, "続": 1059, "綴": 1060, "綿": 1061, "緊": 1062, "総": 1063, "緑": 1064, "緒": 1065, "線": 1066, "編": 1067, "練": 1068, "縁": 1069, "縦": 1070, "縫": 1071, "縮": 1072, "繁": 1073, "繊": 1074, "繋": 1075, "繕": 1076, "纏": 1077, "罅": 1078, "置": 1079, "署": 1080, "羊": 1081, "美": 1082, "群": 1083, "義": 1084, "羽": 1085, "翌": 1086, "習": 1087, "翻": 1088, "老": 1089, "考": 1090, "者": 1091, "耐": 1092, "耳": 1093, "聖": 1094, "聞": 1095, "職": 1096, "肉": 1097, "肋": 1098, "肌": 1099, "肘": 1100, "肝": 1101, "肥": 1102, "肩": 1103, "育": 1104, "胃": 1105, "背": 1106, "胴": 1107, "胸": 1108, "能": 1109, "脅": 1110, "脈": 1111, "脚": 1112, "脱": 1113, "脳": 1114, "腐": 1115, "腕": 1116, "腫": 1117, "腸": 1118, "腹": 1119, "膜": 1120, "膝": 1121, "臓": 1122, "自": 1123, "致": 1124, "興": 1125, "舗": 1126, "舞": 1127, "舟": 1128, "航": 1129, "般": 1130, "舶": 1131, "船": 1132, "艇": 1133, "良": 1134, "色": 1135, "花": 1136, "芸": 1137, "苑": 1138, "苗": 1139, "若": 1140, "苦": 1141, "英": 1142, "茂": 1143, "茶": 1144, "草": 1145, "荷": 1146, "菌": 1147, "菓": 1148, "菜": 1149, "落": 1150, "葉": 1151, "葬": 1152, "蔵": 1153, "薄": 1154, "薬": 1155, "藩": 1156, "虚": 1157, "虜": 1158, "虫": 1159, "融": 1160, "血": 1161, "衆": 1162, "行": 1163, "術": 1164, "街": 1165, "衛": 1166, "衣": 1167, "表": 1168, "袋": 1169, "裂": 1170, "装": 1171, "裏": 1172, "裕": 1173, "補": 1174, "製": 1175, "裾": 1176, "複": 1177, "襞": 1178, "襲": 1179, "西": 1180, "要": 1181, "覆": 1182, "見": 1183, "規": 1184, "視": 1185, "覚": 1186, "親": 1187, "観": 1188, "角": 1189, "解": 1190, "触": 1191, "言": 1192, "計": 1193, "訓": 1194, "記": 1195, "訪": 1196, "設": 1197, "許": 1198, "訳": 1199, "証": 1200, "評": 1201, "試": 1202, "詰": 1203, "話": 1204, "誌": 1205, "認": 1206, "誕": 1207, "語": 1208, "説": 1209, "読": 1210, "誰": 1211, "調": 1212, "談": 1213, "論": 1214, "諸": 1215, "謙": 1216, "講": 1217, "識": 1218, "警": 1219, "議": 1220, "譲": 1221, "護": 1222, "豆": 1223, "豊": 1224, "豚": 1225, "象": 1226, "負": 1227, "財": 1228, "貧": 1229, "販": 1230, "責": 1231, "買": 1232, "貸": 1233, "費": 1234, "貿": 1235, "賀": 1236, "賂": 1237, "賃": 1238, "賄": 1239, "資": 1240, "賑": 1241, "賛": 1242, "賞": 1243, "賢": 1244, "質": 1245, "赤": 1246, "走": 1247, "赴": 1248, "起": 1249, "超": 1250, "越": 1251, "足": 1252, "距": 1253, "跡": 1254, "路": 1255, "踊": 1256, "踏": 1257, "躇": 1258, "躊": 1259, "躍": 1260, "身": 1261, "車": 1262, "軌": 1263, "軍": 1264, "軒": 1265, "転": 1266, "軽": 1267, "輪": 1268, "輸": 1269, "辛": 1270, "辞": 1271, "農": 1272, "辺": 1273, "込": 1274, "迎": 1275, "近": 1276, "返": 1277, "追": 1278, "退": 1279, "送": 1280, "逃": 1281, "逆": 1282, "透": 1283, "途": 1284, "通": 1285, "速": 1286, "造": 1287, "連": 1288, "週": 1289, "進": 1290, "遂": 1291, "遅": 1292, "遇": 1293, "遊": 1294, "運": 1295, "遍": 1296, "過": 1297, "道": 1298, "達": 1299, "違": 1300, "遠": 1301, "適": 1302, "遭": 1303, "選": 1304, "遺": 1305, "避": 1306, "邦": 1307, "邪": 1308, "郊": 1309, "郡": 1310, "部": 1311, "郵": 1312, "郷": 1313, "都": 1314, "配": 1315, "酒": 1316, "酷": 1317, "酸": 1318, "醜": 1319, "里": 1320, "重": 1321, "野": 1322, "量": 1323, "金": 1324, "釣": 1325, "鉄": 1326, "鉛": 1327, "鉢": 1328, "銀": 1329, "銘": 1330, "鋼": 1331, "錨": 1332, "鍵": 1333, "鎮": 1334, "鏡": 1335, "長": 1336, "門": 1337, "閉": 1338, "開": 1339, "間": 1340, "関": 1341, "闇": 1342, "闘": 1343, "阪": 1344, "防": 1345, "降": 1346, "限": 1347, "院": 1348, "除": 1349, "陥": 1350, "陸": 1351, "隊": 1352, "階": 1353, "隔": 1354, "際": 1355, "障": 1356, "隣": 1357, "雄": 1358, "集": 1359, "雑": 1360, "離": 1361, "難": 1362, "雨": 1363, "雪": 1364, "雰": 1365, "雲": 1366, "零": 1367, "電": 1368, "需": 1369, "震": 1370, "霧": 1371, "露": 1372, "青": 1373, "静": 1374, "非": 1375, "面": 1376, "靴": 1377, "鞄": 1378, "音": 1379, "響": 1380, "預": 1381, "頑": 1382, "領": 1383, "頭": 1384, "頼": 1385, "題": 1386, "顔": 1387, "願": 1388, "類": 1389, "顧": 1390, "風": 1391, "飄": 1392, "飛": 1393, "食": 1394, "飯": 1395, "飲": 1396, "飾": 1397, "養": 1398, "餐": 1399, "餓": 1400, "館": 1401, "首": 1402, "馬": 1403, "駅": 1404, "駐": 1405, "騒": 1406, "験": 1407, "骨": 1408, "高": 1409, "髣": 1410, "髪": 1411, "髴": 1412, "魚": 1413, "鮮": 1414, "鯨": 1415, "鰻": 1416, "鳥": 1417, "鳴": 1418, "鵜": 1419, "麗": 1420, "麦": 1421, "黄": 1422, "黒": 1423, "黙": 1424, "鼻": 1425, "齢": 1426, "．": 1427, "？": 1428, "ｄ": 1429, "ｇ": 1430, "ｐ": 1431, "|": 0, "[UNK]": 1432, "[PAD]": 1433}