update

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+vocab/belle_7b_2m/belle-7b-2m/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+vocab/bloom/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,16 @@
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+.idea/

README.md

@@ -1,8 +1,8 @@
 ---
-title: Tokenizer Arena
-emoji: 🚀
-colorFrom: gray
-colorTo: yellow
+title: The Tokenizer Playground
+emoji: ⚡
+colorFrom: red
+colorTo: gray
 sdk: gradio
 sdk_version: 3.41.2
 app_file: app.py

app.py

@@ -0,0 +1,155 @@
+# coding=utf-8
+# author: xusong
+# time: 2022/8/23 16:06
+
+"""
+
+plots
+
+table
+
+## related demo
+http://text-processing.com/demo/tokenize/
+
+## 可视化
+
+[ The, 2, QUICK, Brown, Foxes, jumped, over, the, lazy, dog's, bone ]
+"""
+
+import json
+import pandas as pd
+import gradio as gr
+
+from vocab import all_tokenizers, load_tokener
+
+# 显示空格：https://blog.csdn.net/liuxiao723846/article/details/118994673
+# 隐藏legend：
+css = """
+.space-show {white-space: pre-wrap;}
+.cell-wrap {white-space: pre-wrap;}
+.category-legend {display: none !important}
+"""
+
+example_text = """中文测试：华为智能音箱发布：华为Sound X。維基百科由非營利組織──維基媒體基金會負責維持
+标点测试：，。！？；
+空格测试：  2个空格        8个空格
+数字测试：(10086 + 98) = 100184"""
+
+
+def tokenize(text, tokenizer_type):
+    print(text, tokenizer_type)
+    pos_tokens = []
+    tokenizer = load_tokener(tokenizer_type)
+    encoding = tokenizer.encode(text)
+
+    table = []
+
+    for idx, token_id in enumerate(encoding):
+        decode_text = tokenizer.decode([token_id])  # 特殊字符解码后会统一变成 �，对应 "\ufffd"
+        pos_tokens.extend([(decode_text, str(idx % 3))])
+
+        # token  "Byte":  # 这是 utf-8编码吧？
+        token = tokenizer.convert_ids_to_tokens([token_id])[0]
+        if isinstance(token, bytes):
+            token_str = token.decode("utf-8")
+            token_bytes = token
+            json_dumps = json.dumps(token_str)
+        elif isinstance(token, str):
+            token_str = token
+            token_bytes = bytes(token_str, "utf-8")
+            json_dumps = json.dumps(token_str)
+        else:
+            return
+
+        table.append(
+            {"TokenID": token_id,
+             "Token": token_str,  # utf-8解码后的字符串，为什么有些是 <0xE7>，表示什么？比如llama
+             "Text": decode_text,  #
+             # "Bytes": token_bytes,  # bytes类型在gradio前端页面被解码成字符串，比如   b'\xe4\xb8\xad' 仍然显示成 "中"。因此 str(token_bytes)
+             "Bytes": str(token_bytes),
+             # "Unicode": json_dumps  # unicode, 如果是ascii码，就直接显示。如果不是ascii码，就显示unicode
+             }
+        )
+
+    table_df = pd.DataFrame(table)
+    print(table)
+    print(table_df)
+
+    return pos_tokens, table_df
+
+
+def test_coding():
+    bytes1 = b'\xe4\xb8\xad'
+    print(bytes1)  # b'\xe4\xb8\xad'
+
+
+with gr.Blocks(css=css) as demo:
+    gr.HTML("""<h1 align="center">Tokenizer Arena</h1>""")
+    # links: https://www.coderstool.com/utf8-encoding-decoding
+    #
+
+
+    user_input = gr.Textbox(
+        value=example_text,
+        lines=5
+    )  # placeholder="Enter sentence here..."
+
+    # submitBtn = gr.Button("生成回复", variant="primary")
+
+    # TODO: 图 表 压缩率
+    # llama chatglm gpt_nexo_20b baichuan  baichuan_7b
+    with gr.Row():
+        with gr.Column():
+            tokenizer_type_1 = gr.Dropdown(
+                all_tokenizers, value="llama", label="tokenizer"
+            )
+            token_counter_1 = None  # 计数器
+            output_text_1 = gr.Highlightedtext(
+                label="Tokenization",
+                show_legend=True,
+                elem_classes="space-show"
+            )
+
+            output_table_1 = gr.Dataframe(
+                headers=["TokenID", "Byte", "Text"],
+                datatype=["str", "str", "str"],
+                #elem_classes="space-show",   # 给整个Dataframe加这个css不起作用，因此直接修改cell-wrap
+            )
+
+        with gr.Column():
+            tokenizer_type_2 = gr.Dropdown(
+                all_tokenizers, value="baichuan_7b", label="tokenizer"
+            )
+            token_counter_2 = None  # 计数器
+            output_text_2 = gr.Highlightedtext(
+                label="Tokenization",
+                show_legend=True,
+                elem_classes="space-show"
+            )
+
+            output_table_2 = gr.Dataframe(
+                headers=["TokenID", "Token", "Text"],
+                datatype=["str", "str", "str"],
+            )
+
+        user_input.change(tokenize,
+                          [user_input, tokenizer_type_1],
+                          [output_text_1, output_table_1])
+        tokenizer_type_1.change(tokenize, [user_input, tokenizer_type_1], [output_text_1, output_table_1])
+
+        user_input.change(tokenize,
+                          [user_input, tokenizer_type_2],
+                          [output_text_2, output_table_2])
+        tokenizer_type_2.change(tokenize, [user_input, tokenizer_type_2], [output_text_2, output_table_2])
+
+    # submitBtn.click(tokenize, [user_input, tokenizer_type], outputs,
+    #                 show_progress=True)
+
+    # examples=[
+    #     ["What a beautiful morning for a walk!"],
+    #     ["It was the best of times, it was the worst of times."],
+    #     ["多个空格    It  ss  was the best of times, it was the worst of times."],
+    # ]
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,4 @@
+transformers>=4.21.1
+sentencepiece
+tiktoken
+torch

utils/convert_sp_to_json.py

@@ -0,0 +1,4 @@
+
+from vocab.baichuan_7b import tokenizer
+
+tokenizer.sp

utils/lang_util.py

@@ -0,0 +1,3 @@
+"""
+日语、韩语 等
+"""

utils/oov_util.py

@@ -0,0 +1,12 @@
+
+
+
+import os
+
+CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
+
+space_tokens = ["空格 ，两个空格  ，三个空格    ，制表符\t,换行符\n"]
+
+
+
+docs = [line.strip() for line in open(os.path.join(CURRENT_DIR, "test.txt"), "r", encoding="utf-8")]

utils/symbol.py

@@ -0,0 +1,34 @@
+"""
+special_symbols: https://github.com/google/sentencepiece/blob/master/doc/special_symbols.md
+"""
+
+import sys
+
+
+# 来自 https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/tokenizer/gpt2_tokenization.py#L55
+# 啥意思？
+def bytes_to_unicode():
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a signficant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    _chr = unichr if sys.version_info[0] == 2 else chr
+    bs = list(range(ord("!"), ord("~") + 1)) + list(range(ord("¡"), ord("¬") + 1)) + \
+        list(range(ord("®"), ord("ÿ") + 1))
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8 + n)
+            n += 1
+    cs = [_chr(n) for n in cs]
+    return dict(zip(bs, cs))
+
+aa = bytes_to_unicode()
+print(aa)

utils/text_util.py

@@ -0,0 +1,15 @@
+
+
+
+def is_chinese(uchar):
+    """
+    https://github.com/fxsjy/jieba/blob/master/jieba/__init__.py#L48
+    re.compile("([\u4E00-\u9FD5]+)", re.U)
+    """
+    return u'\u4e00' <= uchar <= u'\u9fa5'
+
+
+
+def has_chinese(text):
+    """ contains Chinese characters """
+    return any(is_chinese(ch) for ch in text)

utils/vocab.jd.txt.v2

@@ -0,0 +1,10268 @@
+[PAD]
+[unused1]
+[unused2]
+[unused3]
+[unused4]
+[unused5]
+[unused6]
+[unused7]
+[unused8]
+[unused9]
+[unused10]
+[unused11]
+[unused12]
+[unused13]
+[unused14]
+[unused15]
+[unused16]
+[unused17]
+[unused18]
+[unused19]
+[unused20]
+[unused21]
+[unused22]
+[unused23]
+[unused24]
+[unused25]
+[unused26]
+[unused27]
+[unused28]
+[unused29]
+[unused30]
+[unused31]
+[unused32]
+[unused33]
+[unused34]
+[unused35]
+[unused36]
+[unused37]
+[unused38]
+[unused39]
+[unused40]
+[unused41]
+[unused42]
+[unused43]
+[unused44]
+[unused45]
+[unused46]
+[unused47]
+[unused48]
+[unused49]
+[unused50]
+[unused51]
+[unused52]
+[unused53]
+[unused54]
+[unused55]
+[unused56]
+[unused57]
+[unused58]
+[unused59]
+[unused60]
+[unused61]
+[unused62]
+[unused63]
+[unused64]
+[unused65]
+[unused66]
+[unused67]
+[unused68]
+[unused69]
+[unused70]
+[unused71]
+[unused72]
+[unused73]
+[unused74]
+[unused75]
+[unused76]
+[unused77]
+[unused78]
+[unused79]
+[unused80]
+[unused81]
+[unused82]
+[unused83]
+[unused84]
+[unused85]
+[unused86]
+[unused87]
+[unused88]
+[unused89]
+[unused90]
+[unused91]
+[unused92]
+[unused93]
+[unused94]
+[unused95]
+[unused96]
+[unused97]
+[unused98]
+[unused99]
+[UNK]
+[CLS]
+[SEP]
+[MASK]
+<S>
+<T>
+!
+"
+”
+“
+—
+–
+…
+’
+‘
+#
+$
+%
+&
+'
+(
+)
+*
++
+,
+-
+.
+/
+0
+1
+2
+3
+4
+5
+6
+7
+8
+9
+10
+11
+12
+13
+14
+15
+16
+17
+18
+19
+20
+21
+22
+23
+24
+25
+26
+27
+28
+29
+30
+31
+32
+33
+34
+35
+36
+37
+38
+39
+40
+41
+42
+43
+44
+45
+46
+47
+48
+49
+50
+51
+52
+53
+54
+55
+56
+57
+58
+59
+60
+61
+62
+63
+64
+65
+66
+67
+68
+69
+70
+71
+72
+73
+74
+75
+76
+77
+78
+79
+80
+81
+82
+83
+84
+85
+86
+87
+88
+89
+90
+91
+92
+93
+94
+95
+96
+97
+98
+99
+100
+120
+128
+180
+200
+256
+304
+360
+500
+512
+1000
+1080
+2000
+2014
+2015
+2016
+2017
+2018
+2019
+2020
+2021
+2022
+:
+;
+<
+=
+>
+?
+@
+[
+\
+]
+^
+_
+a
+b
+c
+d
+e
+f
+g
+h
+i
+j
+k
+l
+m
+n
+o
+p
+q
+r
+s
+t
+u
+v
+w
+x
+y
+z
+{
+|
+}
+~
+£
+¤
+¥
+§
+«
+°
+±
+²
+³
+µ
+·
+¹
+º
+»
+¼
+×
+ß
+æ
+÷
+ø
+đ
+ŋ
+ɔ
+ə
+ɡ
+ʰ
+ˇ
+ˈ
+ˊ
+ˋ
+ˍ
+ː
+˙
+˚
+ˢ
+α
+β
+γ
+δ
+ε
+η
+θ
+ι
+κ
+λ
+μ
+ν
+ο
+π
+ρ
+ς
+σ
+τ
+υ
+φ
+χ
+ψ
+ω
+а
+б
+в
+г
+д
+е
+ж
+з
+и
+к
+л
+м
+н
+о
+п
+р
+с
+т
+у
+ф
+х
+ц
+ч
+ш
+ы
+ь
+я
+і
+ก
+ง
+น
+ม
+ย
+ร
+อ
+า
+เ
+๑
+་
+ღ
+ᵃ
+ᵉ
+ᵍ
+ᵏ
+ᵐ
+ᵒ
+ᵘ
+‖
+„
+†
+•
+‥
+‧
+‰
+′
+″
+‹
+›
+※
+‿
+⁄
+ⁱ
+⁺
+ⁿ
+₁
+₂
+₃
+₄
+€
+℃
+№
+ⅰ
+ⅱ
+ⅲ
+ⅳ
+ⅴ
+←
+↑
+→
+↓
+⇒
+∀
+−
+∕
+∙
+√
+∞
+∟
+∠
+∣
+∥
+∩
+∮
+∶
+∼
+∽
+≈
+≒
+≡
+≤
+≥
+≦
+≧
+≪
+≫
+⊙
+⋅
+⋈
+⋯
+⌒
+①
+②
+③
+④
+⑤
+⑥
+⑦
+⑧
+⑨
+⑩
+⑴
+⑵
+⑶
+⑷
+⑸
+⒈
+⒉
+⒊
+⒋
+ⓒ
+ⓔ
+ⓘ
+─
+━
+│
+┃
+┅
+┆
+┊
+┌
+└
+├
+┣
+═
+║
+╚
+╞
+╠
+╭
+╮
+╯
+╰
+╱
+╳
+▂
+▃
+▅
+▇
+█
+▉
+▋
+▌
+▍
+▎
+■
+□
+▬
+▲
+△
+►
+▼
+▽
+◆
+◇
+○
+◎
+●
+◕
+◠
+◢
+◤
+★
+☆
+☞
+☼
+♡
+♪
+♫
+♬
+✕
+✦
+✪
+✰
+✿
+❀
+➜
+➤
+⦿
+、
+。
+〃
+々
+〇
+〈
+〉
+《
+》
+「
+」
+『
+』
+【
+】
+〓
+〔
+〕
+〖
+〗
+〜
+〝
+〞
+ㄧ
+ㆍ
+㈦
+㊣
+㎡
+㗎
+一
+丁
+七
+万
+丈
+三
+上
+下
+不
+与
+丐
+丑
+专
+且
+丕
+世
+丘
+丙
+业
+丛
+东
+丝
+丞
+両
+丢
+两
+严
+丧
+丨
+个
+丫
+中
+丰
+串
+临
+丶
+丸
+丹
+为
+主
+丼
+丽
+举
+丿
+乂
+乃
+久
+么
+义
+之
+乌
+乍
+乎
+乏
+乐
+乒
+乓
+乔
+乖
+乗
+乘
+乙
+乜
+九
+乞
+也
+习
+乡
+书
+乩
+买
+乱
+乳
+亀
+了
+予
+争
+事
+二
+于
+亏
+云
+互
+五
+井
+亘
+亚
+些
+亜
+亟
+亡
+亢
+交
+亥
+亦
+产
+亨
+亩
+享
+京
+亭
+亮
+亲
+亳
+亵
+人
+亿
+什
+仁
+仃
+仄
+仅
+仆
+仇
+今
+介
+仍
+从
+仏
+仑
+仓
+仔
+仕
+他
+仗
+付
+仙
+仝
+仞
+仟
+代
+令
+以
+仨
+仪
+们
+仮
+仰
+仲
+件
+价
+任
+份
+仿
+企
+伉
+伊
+伍
+伎
+伏
+伐
+休
+伕
+众
+优
+伙
+会
+伝
+伞
+伟
+传
+伢
+伤
+伦
+伪
+伫
+伯
+估
+伴
+伶
+伸
+伺
+似
+伽
+佃
+但
+位
+低
+住
+佐
+佑
+体
+佔
+何
+佗
+佘
+余
+佚
+佛
+作
+佝
+佞
+佟
+你
+佢
+佣
+佤
+佥
+佩
+佬
+佯
+佰
+佳
+佶
+佻
+佼
+使
+侃
+侄
+侈
+例
+侍
+侏
+侑
+侗
+供
+依
+侠
+価
+侣
+侥
+侦
+侧
+侨
+侬
+侮
+侯
+侵
+侷
+便
+促
+俄
+俊
+俎
+俏
+俐
+俑
+俗
+俘
+俚
+保
+俞
+俟
+信
+俨
+俩
+俪
+俬
+俭
+修
+俯
+俱
+俳
+俸
+俺
+俾
+倌
+倍
+倏
+倒
+倔
+倖
+倘
+候
+倚
+倜
+借
+倡
+値
+倦
+倩
+倪
+倬
+倭
+倶
+债
+值
+倾
+偃
+假
+偈
+偌
+偎
+偏
+偕
+做
+停
+健
+偶
+偷
+偻
+偿
+傀
+傅
+傍
+傚
+傣
+傥
+储
+傩
+催
+傲
+傻
+働
+像
+僖
+僚
+僧
+僭
+僮
+僱
+僵
+僻
+儆
+儋
+儒
+儡
+儿
+兀
+允
+元
+兄
+充
+兆
+先
+光
+克
+免
+児
+兑
+兔
+兖
+党
+兜
+兢
+入
+全
+八
+公
+六
+兮
+兰
+共
+兲
+关
+兴
+兵
+其
+具
+典
+兹
+养
+兼
+兽
+冀
+内
+円
+冇
+冈
+冉
+册
+再
+冏
+��
+冕
+冗
+写
+军
+农
+冠
+冢
+冤
+冥
+冨
+冬
+冯
+冰
+冲
+决
+况
+冶
+冷
+冻
+冼
+冽
+冾
+净
+凄
+准
+凇
+凉
+凋
+凌
+减
+凑
+凛
+凝
+几
+凡
+凤
+処
+凪
+凭
+凯
+凰
+凳
+凶
+凸
+凹
+出
+击
+函
+凿
+刀
+刁
+刃
+分
+切
+刈
+刊
+刍
+刎
+刑
+划
+列
+刘
+则
+刚
+创
+初
+删
+判
+刨
+利
+别
+刮
+到
+制
+刷
+券
+刹
+刺
+刻
+刽
+剁
+剂
+剃
+剉
+削
+剌
+前
+剐
+剑
+剔
+剖
+剜
+剣
+剤
+剥
+剧
+剩
+剪
+副
+割
+剷
+剽
+剿
+劈
+力
+劝
+办
+功
+加
+务
+劣
+动
+助
+努
+劫
+劭
+励
+劲
+劳
+労
+劵
+効
+劾
+势
+勃
+勇
+勉
+勋
+勐
+勒
+勖
+勘
+募
+勤
+勧
+勳
+勺
+勾
+勿
+匀
+包
+匆
+匈
+匍
+匐
+匕
+化
+北
+匙
+匝
+匠
+匡
+匣
+匪
+匮
+匹
+区
+医
+匾
+匿
+十
+千
+卅
+升
+午
+卉
+半
+卍
+华
+协
+卑
+卒
+卓
+单
+卖
+南
+単
+博
+卜
+卞
+卟
+占
+卡
+卢
+卤
+卦
+卧
+卫
+卮
+卯
+印
+危
+即
+却
+卵
+卷
+卸
+卿
+厂
+厄
+厅
+历
+厉
+压
+厌
+厕
+厘
+厚
+厝
+原
+厢
+厥
+厦
+厨
+厩
+厮
+厳
+去
+县
+叁
+参
+又
+叉
+及
+友
+双
+反
+収
+发
+叔
+取
+受
+变
+叙
+叛
+叟
+叠
+叡
+口
+古
+句
+另
+叨
+叩
+只
+叫
+召
+叭
+叮
+可
+台
+叱
+史
+右
+叵
+叶
+号
+司
+叹
+叻
+叼
+叽
+吁
+吃
+各
+吆
+合
+吉
+吊
+吋
+同
+名
+后
+吏
+吐
+向
+吒
+吓
+吕
+吖
+吗
+君
+吝
+吞
+吟
+吠
+吡
+否
+吧
+吨
+吩
+含
+听
+吭
+吮
+启
+吱
+吴
+吵
+吸
+吹
+吻
+吼
+吽
+吾
+呀
+呃
+呆
+呈
+告
+呋
+呎
+呐
+呓
+呕
+呗
+员
+呛
+呜
+呢
+呤
+呦
+周
+呱
+呲
+味
+呵
+呷
+呸
+呻
+呼
+命
+咀
+咁
+咂
+咄
+咆
+咋
+和
+咎
+咏
+咐
+咒
+咔
+咕
+咖
+咗
+咘
+咙
+咚
+咛
+咣
+咤
+咦
+咧
+咨
+咩
+咪
+咫
+咬
+咭
+咯
+咱
+咲
+咳
+咸
+咻
+咽
+咿
+哀
+品
+哂
+哄
+哆
+哇
+哈
+哉
+哋
+哌
+响
+哎
+哏
+哐
+哑
+哒
+哔
+哗
+哟
+哥
+哦
+哧
+哨
+哩
+哪
+哭
+哮
+哲
+哺
+哼
+哽
+唁
+唆
+唇
+唉
+唏
+唐
+唑
+唔
+唠
+唤
+唧
+唬
+售
+唯
+唰
+唱
+唳
+唷
+唸
+唾
+啃
+啄
+商
+啉
+啊
+啕
+啖
+啜
+啡
+啤
+啥
+啦
+啧
+啪
+啫
+啬
+啮
+啰
+啱
+啲
+啵
+啶
+啷
+啸
+啻
+啼
+啾
+喀
+喂
+喃
+善
+喆
+喇
+喉
+喊
+喋
+喏
+喔
+喘
+喙
+喜
+喝
+喟
+喧
+喫
+喰
+喱
+喳
+喵
+営
+喷
+喹
+喺
+喻
+喽
+嗅
+嗑
+嗒
+嗓
+嗔
+嗖
+嗜
+嗝
+嗟
+嗡
+嗣
+嗤
+嗦
+嗨
+嗪
+嗬
+嗯
+嗰
+嗲
+嗳
+嗷
+嗽
+嘀
+嘅
+嘈
+嘉
+嘌
+嘎
+嘘
+嘚
+嘛
+嘞
+嘟
+嘢
+嘣
+嘤
+嘧
+嘭
+嘱
+嘲
+嘴
+嘶
+嘹
+嘻
+嘿
+噌
+噎
+噔
+噗
+噙
+噜
+噢
+噤
+器
+噩
+噪
+噬
+噱
+噶
+噻
+噼
+嚎
+嚏
+嚐
+嚓
+嚟
+嚣
+嚷
+嚼
+囉
+囊
+囍
+囔
+囗
+囚
+四
+囝
+回
+囟
+因
+囡
+团
+団
+囤
+囧
+囫
+园
+困
+囱
+囲
+図
+围
+囹
+固
+国
+图
+囿
+圃
+圄
+圆
+圈
+圏
+圜
+土
+圣
+圧
+在
+圩
+圭
+地
+圳
+场
+圻
+圾
+址
+坂
+均
+坊
+坍
+坎
+坏
+坐
+坑
+块
+坚
+坛
+坝
+坞
+坟
+坠
+坡
+坤
+坦
+坨
+坪
+坯
+坳
+坵
+坷
+垂
+垃
+垄
+型
+垒
+垚
+垛
+垠
+垢
+垣
+垦
+垩
+垫
+垭
+垮
+埂
+埃
+埋
+城
+埔
+埕
+埗
+域
+埠
+埤
+埵
+埸
+培
+基
+埼
+堀
+堂
+堃
+堆
+堇
+堑
+堕
+堙
+堡
+堤
+堪
+堰
+堵
+堺
+堿
+塌
+塑
+塔
+塘
+塞
+塩
+填
+塬
+塭
+塾
+墀
+境
+墅
+墉
+墒
+墓
+増
+墘
+墙
+增
+墟
+墨
+墩
+壁
+壅
+壆
+壊
+壑
+壕
+壤
+士
+壬
+壮
+声
+売
+壳
+壶
+壹
+处
+备
+変
+复
+夏
+夔
+夕
+外
+夙
+多
+夜
+够
+夥
+大
+天
+太
+夫
+夭
+央
+夯
+失
+头
+夷
+夸
+夹
+夺
+奂
+奄
+奇
+奈
+奉
+奋
+奎
+奏
+契
+奔
+奕
+奖
+套
+奘
+奚
+奠
+奢
+奥
+女
+奴
+奶
+奸
+她
+好
+如
+妃
+妄
+妆
+妇
+妈
+妊
+妍
+妒
+妓
+妖
+妘
+妙
+妞
+妣
+妤
+妥
+妨
+妩
+妪
+妮
+妲
+妳
+妹
+妻
+妾
+姆
+姉
+姊
+始
+姐
+姑
+姒
+姓
+委
+姗
+姚
+姜
+姝
+姣
+姥
+姨
+姪
+姫
+姬
+姹
+姻
+姿
+威
+娃
+娄
+娅
+娆
+娇
+娉
+娑
+娓
+娘
+娜
+娟
+娠
+娣
+娥
+娩
+娱
+娲
+娴
+娶
+娼
+婀
+婆
+婉
+婊
+婕
+婚
+婢
+婧
+婪
+婴
+婵
+婶
+婷
+婺
+婿
+媒
+媚
+媛
+媞
+媲
+媳
+媾
+嫁
+嫂
+嫉
+嫌
+嫑
+嫔
+嫖
+嫘
+嫚
+嫡
+嫣
+嫦
+嫩
+嫲
+嬅
+嬉
+嬗
+嬛
+嬢
+嬴
+嬷
+嬿
+孀
+孃
+子
+孑
+孔
+孕
+孖
+字
+存
+孙
+孚
+孛
+孜
+孝
+孟
+孢
+季
+孤
+学
+孩
+孪
+孬
+孰
+孱
+孳
+孵
+孺
+孽
+宁
+它
+宅
+宇
+守
+安
+宋
+完
+宏
+宓
+宕
+宗
+官
+宙
+定
+宛
+宜
+宝
+实
+実
+宠
+审
+客
+宣
+室
+宥
+宦
+宪
+宫
+宰
+害
+宴
+宵
+家
+宸
+容
+宽
+宾
+宿
+寂
+寄
+寅
+密
+寇
+富
+寐
+寒
+寓
+寛
+寝
+寞
+察
+寡
+寥
+寨
+寮
+寰
+寸
+对
+寺
+寻
+导
+対
+寿
+封
+専
+射
+将
+尉
+尊
+小
+少
+尔
+尕
+尖
+尘
+尚
+尝
+尤
+尧
+尬
+就
+尴
+尸
+尹
+尺
+尻
+尼
+尽
+尾
+尿
+局
+屁
+层
+屄
+居
+屈
+屉
+届
+屋
+屌
+屎
+屏
+屐
+屑
+展
+属
+屠
+屡
+履
+屯
+山
+屹
+屿
+岀
+岁
+岂
+岌
+岐
+岑
+岔
+岖
+岗
+岘
+岙
+岚
+岛
+岩
+岫
+岬
+岭
+岱
+岳
+岷
+岸
+��
+峋
+峒
+峙
+峡
+峤
+峥
+峦
+峨
+峪
+峭
+峯
+峰
+峻
+崁
+崂
+崆
+崇
+崎
+崑
+崔
+崖
+崙
+崛
+崧
+崩
+崭
+崴
+崽
+嵇
+嵊
+嵋
+嵌
+嵘
+嵩
+嵬
+嵯
+嶂
+嶋
+嶙
+巅
+巍
+巖
+川
+州
+巡
+巢
+工
+左
+巧
+巨
+巩
+巫
+差
+己
+已
+巳
+巴
+巷
+巻
+巽
+巾
+巿
+币
+市
+布
+帅
+帆
+师
+希
+帐
+帑
+帕
+帖
+帘
+帚
+帛
+帜
+帝
+带
+帧
+席
+帮
+帯
+帰
+帷
+常
+帼
+帽
+幂
+幄
+幅
+幌
+幔
+幕
+幡
+幢
+干
+平
+年
+并
+幸
+幻
+幼
+幽
+广
+庁
+広
+庄
+庆
+庇
+床
+序
+庐
+库
+应
+底
+庖
+店
+庙
+庚
+府
+庞
+废
+庠
+度
+座
+庭
+庵
+庶
+康
+庸
+庹
+庾
+廃
+廉
+廊
+廓
+廖
+延
+廷
+建
+廿
+开
+弁
+异
+弃
+弄
+弈
+弊
+弋
+式
+弑
+弓
+弔
+引
+弗
+弘
+弛
+弟
+张
+弥
+弦
+弧
+弩
+弭
+弯
+弱
+弹
+强
+弼
+弾
+彅
+归
+当
+录
+彗
+彝
+形
+彤
+彦
+彧
+彩
+彪
+彫
+彬
+彭
+彰
+影
+彷
+役
+彻
+彼
+彿
+往
+征
+径
+待
+徇
+很
+徉
+徊
+律
+徐
+徒
+従
+徕
+得
+徘
+徙
+徜
+御
+徨
+循
+徬
+微
+徳
+徴
+德
+徼
+徽
+心
+必
+忆
+忌
+忍
+忏
+忐
+忑
+忒
+忖
+志
+忘
+忙
+応
+忠
+忡
+忤
+忧
+忪
+快
+忱
+念
+忻
+忽
+忿
+怀
+态
+怂
+怅
+怆
+怎
+怏
+怒
+怔
+怕
+怖
+怙
+怜
+思
+怠
+怡
+急
+怦
+性
+怨
+怪
+怯
+怵
+总
+怼
+恁
+恃
+恋
+恍
+恐
+恒
+恕
+恙
+恚
+恢
+恣
+恤
+恨
+恩
+恪
+恫
+恬
+恭
+息
+恰
+恳
+恵
+恶
+恸
+恺
+恻
+恼
+恿
+悄
+悉
+悌
+悍
+悔
+悖
+悚
+悟
+悠
+患
+悦
+您
+悩
+悪
+悬
+悯
+悱
+悲
+悴
+悸
+悻
+悼
+悽
+情
+惆
+惇
+惊
+惋
+惑
+惕
+惘
+惚
+惜
+惟
+惠
+惦
+惧
+惨
+惩
+惫
+惬
+惭
+惮
+惯
+惰
+想
+惴
+惶
+惹
+惺
+愁
+愆
+愈
+愉
+愍
+意
+愕
+愚
+感
+愣
+愤
+愧
+愫
+愿
+慈
+慌
+慎
+慑
+慕
+慢
+慧
+慨
+慰
+慵
+慷
+慾
+憋
+憎
+憔
+憧
+憨
+憩
+憬
+憾
+懂
+懈
+懊
+懋
+懑
+懒
+懦
+懵
+懿
+戈
+戊
+戌
+戍
+戎
+戏
+成
+我
+戒
+戕
+或
+战
+戚
+戛
+戟
+戡
+戦
+截
+戬
+戮
+戳
+戴
+户
+戸
+戻
+戾
+房
+所
+扁
+扇
+扈
+扉
+手
+才
+扎
+扑
+扒
+打
+扔
+払
+托
+扛
+扣
+扦
+执
+扩
+扪
+扫
+扬
+扭
+扮
+扯
+扰
+扱
+扳
+扶
+批
+扼
+找
+承
+技
+抄
+抉
+把
+抑
+抒
+抓
+投
+抖
+抗
+折
+抚
+抛
+抜
+択
+抟
+抠
+抡
+抢
+护
+报
+抨
+披
+抬
+抱
+抵
+抹
+押
+抽
+抿
+拂
+拄
+担
+拆
+拇
+拈
+拉
+拌
+拍
+拎
+拐
+拒
+拓
+拔
+拖
+拗
+拘
+拙
+拚
+招
+拜
+拟
+拡
+拢
+拣
+拥
+拦
+拧
+拨
+择
+括
+拭
+拮
+拯
+拱
+拳
+拴
+拷
+拼
+拽
+拾
+拿
+持
+挂
+指
+挈
+按
+挎
+挑
+挖
+挙
+挚
+挛
+挝
+挞
+挟
+挠
+挡
+挣
+挤
+挥
+挨
+挪
+挫
+振
+挲
+挹
+挺
+挽
+捂
+捅
+捆
+捉
+捋
+捌
+捍
+捎
+捏
+捐
+捕
+捞
+损
+捡
+换
+捣
+捧
+捩
+据
+捱
+捲
+捶
+捷
+捺
+捻
+掀
+掂
+掇
+授
+掉
+掌
+掏
+掐
+排
+掖
+掘
+掠
+探
+掣
+接
+控
+推
+掩
+措
+掬
+掰
+掲
+掳
+掴
+掷
+掸
+掺
+揃
+揄
+揆
+揉
+揍
+描
+提
+插
+揖
+握
+揣
+揩
+揪
+揭
+援
+揶
+揸
+揹
+揽
+搀
+搁
+搂
+搅
+搏
+搐
+搓
+搔
+搜
+搞
+搡
+搪
+搬
+搭
+携
+搽
+摀
+摁
+摄
+摆
+摇
+摈
+摊
+摒
+摔
+摘
+摞
+摧
+摩
+摸
+摹
+撂
+撃
+撅
+撇
+撑
+撒
+撕
+撚
+撞
+撤
+撩
+撬
+播
+撮
+撰
+撵
+撷
+撸
+撼
+擀
+擂
+擅
+操
+擎
+擒
+擘
+擞
+擡
+擢
+擦
+攀
+攒
+攘
+攞
+攥
+攫
+支
+收
+攸
+改
+攻
+放
+政
+故
+效
+敌
+敍
+敎
+敏
+救
+敕
+敖
+教
+敛
+敝
+敞
+敢
+散
+敦
+敬
+数
+敲
+整
+敷
+文
+斋
+斌
+斎
+斐
+斑
+斓
+斗
+料
+斛
+斜
+斟
+斡
+斤
+斥
+斧
+斩
+斫
+断
+斯
+新
+方
+施
+旁
+旃
+旅
+旋
+旌
+旎
+族
+旖
+旗
+无
+既
+日
+旦
+旧
+旨
+早
+旬
+旭
+旮
+旱
+时
+旷
+旺
+旻
+昀
+昂
+昆
+昇
+昉
+昊
+昌
+明
+昏
+易
+昔
+昕
+昙
+星
+映
+春
+昧
+昨
+昭
+是
+昱
+昴
+昵
+昶
+昼
+显
+晁
+晃
+晋
+晌
+晏
+晒
+晓
+晔
+晕
+晖
+晗
+晚
+晞
+晟
+晤
+晦
+晨
+晩
+普
+景
+晰
+晴
+晶
+晷
+智
+晾
+暂
+暄
+暇
+暌
+暐
+暑
+暖
+暗
+暝
+暧
+暨
+暮
+暱
+暴
+暸
+暹
+曙
+曜
+曝
+曦
+曰
+曲
+曳
+更
+曹
+曼
+曾
+替
+最
+月
+有
+朋
+服
+朐
+朔
+朕
+朗
+望
+朝
+期
+朦
+木
+未
+末
+本
+札
+术
+朱
+朴
+朵
+机
+朽
+杀
+杂
+权
+杆
+杈
+杉
+李
+杏
+材
+村
+杓
+杖
+杜
+杞
+束
+杠
+条
+来
+杨
+杭
+杯
+杰
+杳
+杵
+杷
+杼
+松
+板
+极
+构
+枇
+枉
+枋
+析
+枕
+林
+枚
+果
+枝
+枢
+枣
+枪
+枫
+枭
+枯
+枰
+枱
+枳
+架
+枷
+枸
+柄
+柏
+某
+柑
+柒
+染
+柔
+柘
+柚
+柜
+柞
+柠
+柢
+查
+柩
+柬
+柯
+柱
+柳
+柴
+査
+柿
+栀
+栃
+栄
+栅
+标
+栈
+栉
+栋
+栎
+栏
+树
+栓
+栖
+栗
+校
+栩
+株
+样
+核
+根
+格
+栽
+栾
+桀
+桁
+桂
+桃
+桅
+框
+案
+桉
+桌
+桎
+桐
+桑
+桓
+桔
+桜
+桠
+桡
+桢
+档
+桥
+桦
+桧
+桨
+桩
+桶
+梁
+梅
+梆
+梏
+梓
+梗
+梢
+梦
+梧
+梨
+梭
+梯
+械
+梳
+梵
+梶
+检
+棂
+棉
+棋
+棍
+棒
+棕
+棘
+棚
+棠
+棣
+森
+棱
+棵
+棹
+棺
+椁
+椅
+椋
+��
+椎
+椒
+検
+椪
+椭
+椰
+椹
+椽
+椿
+楂
+楔
+楚
+楝
+楞
+楠
+楣
+楫
+楮
+楷
+楸
+楹
+楼
+楽
+概
+榄
+榆
+榈
+榉
+榔
+榕
+榖
+榛
+榜
+榨
+榫
+榭
+榱
+榴
+榷
+榻
+槁
+槃
+槌
+槎
+槐
+槓
+様
+槛
+槟
+槭
+槲
+槻
+槽
+槿
+樊
+樑
+樟
+模
+権
+横
+樫
+樯
+樱
+樵
+樽
+樾
+橄
+橇
+橐
+橘
+橙
+橡
+橱
+橹
+橼
+檀
+檄
+檎
+檐
+檗
+檬
+欠
+次
+欢
+欣
+欧
+欲
+欸
+欺
+款
+歆
+歇
+歉
+歌
+歎
+歓
+歙
+歛
+止
+正
+此
+步
+武
+歧
+歩
+歪
+歯
+歳
+歴
+歹
+死
+歼
+殁
+殃
+殆
+殇
+殉
+殊
+残
+殒
+殓
+殖
+殡
+殭
+殴
+段
+殷
+殿
+毁
+毂
+毅
+毋
+母
+毎
+每
+毒
+毓
+比
+毕
+毗
+毘
+毙
+毛
+毡
+毫
+毯
+毽
+氏
+氐
+民
+氓
+气
+氖
+気
+氙
+氛
+氟
+氡
+氢
+氤
+氦
+氧
+氨
+氪
+氮
+氯
+氰
+氲
+水
+氷
+永
+氹
+氾
+汀
+汁
+求
+汆
+汇
+汉
+汎
+汐
+汕
+汗
+汛
+汝
+汞
+江
+池
+污
+汤
+汨
+汩
+汪
+汰
+汲
+汴
+汶
+汹
+汽
+汾
+沁
+沂
+沃
+沅
+沈
+沉
+沌
+沏
+沐
+沓
+沙
+沛
+沟
+没
+沢
+沣
+沥
+沦
+沧
+沪
+沫
+沭
+沮
+沱
+河
+沸
+油
+治
+沼
+沽
+沾
+沿
+泄
+泉
+泊
+泌
+泓
+法
+泗
+泛
+泞
+泠
+泡
+波
+泣
+泥
+注
+泪
+泫
+泮
+泯
+泰
+泱
+泳
+泵
+泷
+泸
+泻
+泼
+泽
+泾
+洁
+洄
+洋
+洒
+洗
+洙
+洛
+洞
+津
+洩
+洪
+洮
+洱
+洲
+洵
+洸
+洹
+活
+洼
+洽
+派
+流
+浃
+浄
+浅
+浆
+浇
+浊
+测
+济
+浏
+浑
+浒
+浓
+浔
+浙
+浚
+浜
+浣
+浦
+浩
+浪
+浬
+浮
+浯
+浴
+海
+浸
+涂
+涅
+消
+涉
+涌
+涎
+涓
+涔
+涕
+涙
+涛
+涝
+涞
+涟
+涠
+涡
+涣
+涤
+润
+涧
+涨
+涩
+涪
+涮
+涯
+液
+涵
+涸
+涿
+淀
+淄
+淅
+淆
+淇
+淋
+淌
+淑
+淖
+淘
+淙
+淞
+淡
+淤
+淦
+淫
+淬
+淮
+深
+淳
+混
+淹
+添
+淼
+清
+済
+渉
+渊
+渋
+渍
+渎
+渐
+渔
+渗
+渚
+渝
+渠
+渡
+渣
+渤
+渥
+温
+渭
+港
+渲
+渴
+游
+渺
+湃
+湄
+湍
+湖
+湘
+湛
+湟
+湧
+湫
+湮
+湳
+湾
+湿
+満
+溃
+溅
+溉
+溏
+源
+溜
+溟
+溢
+溥
+溧
+溪
+溯
+溱
+溴
+溶
+溺
+溼
+滁
+滂
+滇
+滋
+滑
+滓
+滔
+滕
+滙
+滚
+滝
+滞
+滟
+满
+滢
+滤
+滥
+滦
+滨
+滩
+滴
+漂
+漆
+漉
+漏
+漓
+演
+漕
+漠
+漩
+漪
+漫
+漯
+漱
+漳
+漾
+潆
+潇
+潋
+潍
+潘
+潜
+潞
+潟
+潢
+潦
+潧
+潭
+潮
+潴
+潸
+潺
+潼
+澄
+澈
+澍
+澎
+澜
+澡
+澧
+澳
+澹
+激
+濂
+濑
+濒
+濠
+濡
+濬
+濮
+濯
+瀑
+瀚
+瀛
+瀞
+瀬
+灌
+灏
+灞
+火
+灬
+灭
+灯
+灰
+灵
+灶
+灸
+灼
+灾
+灿
+炀
+炁
+炅
+炉
+炊
+炎
+炒
+炔
+炕
+炖
+炙
+炜
+炫
+炬
+炭
+炮
+炯
+炳
+炷
+炸
+点
+炼
+炽
+烁
+烂
+烃
+烈
+烊
+烘
+烙
+烛
+烟
+烤
+烦
+烧
+烨
+烩
+烫
+烬
+热
+烯
+烷
+烹
+烽
+焉
+焊
+焕
+焖
+焗
+焘
+焙
+焚
+焜
+焦
+焯
+焰
+焱
+然
+焼
+煅
+煊
+煌
+煎
+煖
+煜
+煞
+煤
+煦
+照
+煨
+煮
+煲
+煸
+煽
+熄
+熊
+熏
+熔
+熙
+熟
+熠
+熨
+熬
+熵
+熹
+燃
+燄
+燊
+燎
+燔
+燕
+燥
+燧
+燮
+燻
+燿
+爆
+爪
+爬
+爰
+爱
+爵
+父
+爷
+爸
+爹
+爻
+爽
+片
+版
+牌
+牍
+牒
+牙
+牛
+牝
+牟
+牠
+牡
+牢
+牦
+牧
+物
+牯
+牲
+牴
+牵
+特
+牺
+犀
+犁
+犄
+犊
+犍
+犒
+犬
+犯
+状
+犷
+犸
+犹
+狂
+狄
+狈
+狎
+狐
+狒
+狗
+狙
+狞
+狠
+狡
+狩
+独
+狭
+狮
+狰
+狱
+狸
+狼
+猎
+猕
+猖
+猗
+猛
+猜
+猝
+猥
+猩
+猪
+猫
+猬
+献
+猴
+猷
+猾
+猿
+獐
+獒
+獗
+獠
+獣
+獭
+獾
+玄
+率
+玉
+王
+玑
+玖
+玛
+玟
+玠
+玥
+玩
+玫
+玮
+环
+现
+玲
+玳
+玷
+玺
+玻
+珀
+珂
+珅
+珈
+珉
+珊
+珍
+珏
+珐
+珑
+珙
+珞
+珠
+珣
+珥
+珩
+珪
+班
+珮
+珲
+珺
+球
+琅
+理
+琇
+琉
+琊
+琍
+琏
+琐
+琛
+琢
+琥
+琦
+琨
+琪
+琬
+琮
+琰
+琲
+琳
+琴
+琵
+琶
+琼
+瑀
+瑁
+瑄
+瑕
+瑗
+瑙
+瑚
+瑛
+瑜
+瑞
+瑟
+瑠
+瑯
+瑰
+瑶
+瑾
+璀
+璁
+璃
+璇
+璋
+璎
+璐
+璜
+璞
+璟
+璧
+璨
+璿
+瓒
+瓜
+瓢
+瓣
+瓤
+瓦
+瓮
+瓯
+瓴
+瓶
+瓷
+甄
+甕
+甘
+甙
+甚
+甜
+生
+甥
+甦
+用
+甩
+甫
+甬
+甭
+甯
+田
+由
+甲
+申
+电
+男
+甸
+町
+画
+甾
+畀
+畅
+界
+畏
+畑
+畔
+留
+畜
+略
+畦
+番
+畲
+畳
+畴
+畸
+畹
+畿
+疆
+疏
+疑
+疔
+疖
+疗
+疙
+疚
+疝
+疟
+疡
+疣
+疤
+疥
+疫
+疮
+疯
+疱
+疲
+疳
+疵
+疸
+疹
+疼
+疽
+疾
+痂
+病
+症
+痈
+痉
+痊
+痍
+痒
+痔
+痕
+痘
+痛
+痞
+痠
+痢
+痣
+痤
+痧
+痨
+痪
+痫
+痰
+痱
+痴
+痹
+痺
+痼
+痿
+瘀
+瘁
+瘘
+瘙
+瘟
+瘠
+瘢
+瘤
+瘦
+瘩
+瘪
+瘫
+瘴
+瘸
+瘾
+癌
+癒
+癖
+癜
+癞
+癡
+癣
+癫
+癸
+発
+登
+白
+百
+皂
+的
+皆
+皇
+皈
+皋
+皎
+皑
+皓
+皖
+皙
+皮
+皱
+皴
+皿
+盂
+盅
+盆
+盈
+益
+盎
+盏
+盐
+监
+盒
+盔
+盖
+盗
+盘
+盛
+盟
+盥
+目
+盯
+盱
+盲
+直
+相
+盹
+盼
+盾
+省
+眈
+眉
+看
+県
+眙
+眞
+真
+眠
+眦
+眨
+眩
+眯
+眶
+眷
+眸
+眺
+眼
+着
+睁
+睇
+睐
+睑
+睛
+睡
+睢
+督
+睥
+睦
+睨
+睪
+睫
+睬
+睹
+睽
+睾
+睿
+瞄
+瞅
+瞇
+瞋
+瞌
+瞎
+瞑
+瞒
+瞓
+瞟
+瞠
+瞥
+瞧
+瞩
+瞪
+瞬
+瞰
+瞳
+瞻
+瞿
+矍
+矗
+矛
+矜
+矢
+��
+知
+矩
+矫
+短
+矮
+石
+矶
+矽
+矾
+矿
+码
+砂
+砌
+砍
+砒
+研
+砖
+砗
+砚
+砝
+砣
+砥
+砧
+砭
+砰
+砲
+破
+砷
+砸
+砺
+砼
+砾
+础
+硅
+硐
+硒
+硕
+硝
+硫
+硬
+确
+硼
+碁
+碇
+碉
+碌
+碍
+碎
+碑
+碓
+碗
+碘
+碚
+碛
+碟
+碣
+碧
+碰
+碱
+碳
+碴
+碾
+磁
+磅
+磊
+磋
+磐
+磕
+磡
+磨
+磬
+磲
+磷
+磺
+礁
+礡
+礴
+示
+礼
+社
+祀
+祁
+祂
+祇
+祈
+祉
+祎
+祐
+祕
+祖
+祗
+祚
+祛
+祜
+祝
+神
+祟
+祠
+祢
+祥
+票
+祭
+祯
+祷
+祸
+祺
+禀
+禁
+禄
+禅
+福
+禛
+禧
+禹
+禺
+离
+禽
+禾
+秀
+私
+秃
+秆
+秉
+秋
+种
+科
+秒
+秘
+租
+秣
+秤
+秦
+秧
+秩
+秭
+积
+称
+秸
+移
+秽
+稀
+程
+稍
+税
+稔
+稗
+稚
+稜
+稞
+稠
+稣
+稲
+稳
+稷
+稹
+稻
+稼
+稽
+稿
+穂
+穆
+穗
+穴
+究
+穷
+穹
+空
+穿
+突
+窃
+窄
+窈
+窍
+窑
+窒
+窓
+窕
+窖
+窗
+窘
+窜
+窝
+窟
+窠
+窥
+窦
+窨
+窿
+立
+竖
+站
+竜
+竞
+竟
+章
+竣
+童
+竭
+端
+竹
+竺
+竽
+竿
+笃
+笆
+笈
+笋
+笏
+笑
+笔
+笙
+笛
+笞
+笠
+符
+笨
+第
+笹
+笺
+笼
+等
+筊
+筋
+筏
+筐
+筑
+筒
+答
+策
+筛
+筝
+筠
+筱
+筲
+筵
+筷
+筹
+签
+简
+箇
+箍
+箐
+箔
+箕
+算
+箝
+管
+箩
+箫
+箭
+箱
+箴
+箸
+篁
+篆
+篇
+篑
+篓
+篙
+篝
+篠
+篡
+篪
+篮
+篱
+篷
+簇
+簌
+簦
+簧
+簪
+簷
+簸
+簿
+籁
+籍
+籐
+籤
+米
+类
+籼
+籽
+粄
+粉
+粑
+粒
+粕
+粗
+粘
+粟
+粤
+粥
+粧
+粪
+粮
+粱
+粲
+粳
+粹
+粼
+粽
+精
+粿
+糅
+糊
+糍
+糕
+糖
+糗
+糙
+糜
+糟
+糠
+糬
+糯
+糸
+系
+紊
+素
+索
+紧
+紫
+紮
+累
+絃
+経
+絮
+絵
+綑
+継
+続
+綦
+総
+縁
+縄
+繁
+繇
+繋
+纂
+纠
+红
+纣
+纤
+约
+级
+纨
+纪
+纫
+纬
+纭
+纯
+纰
+纱
+纲
+纳
+纵
+纶
+纷
+纸
+纹
+纺
+纽
+纾
+线
+绀
+练
+组
+绅
+细
+织
+终
+绊
+绍
+绎
+经
+绑
+绒
+结
+绔
+绕
+绘
+给
+绚
+绛
+络
+绝
+绞
+统
+绡
+绢
+绣
+绥
+绦
+继
+绩
+绪
+绫
+续
+绮
+绯
+绰
+绳
+维
+绵
+绶
+绷
+绸
+绻
+综
+绽
+绾
+绿
+缀
+缄
+缅
+缆
+缇
+缈
+缉
+缎
+缓
+缔
+缕
+编
+缘
+缙
+缚
+缜
+缝
+缠
+缢
+缤
+缥
+缨
+缩
+缪
+缭
+缮
+缰
+缱
+缴
+缸
+缺
+罂
+罄
+罐
+网
+罔
+罕
+罗
+罚
+罡
+罢
+罩
+罪
+置
+署
+罹
+羁
+羊
+羌
+美
+羔
+羚
+羞
+羟
+羡
+羣
+群
+羧
+羨
+羯
+羲
+羸
+羹
+羽
+羿
+翁
+翅
+翊
+翌
+翎
+翔
+翘
+翟
+翠
+翡
+翦
+翩
+翰
+翱
+翳
+翻
+翼
+耀
+老
+考
+耄
+者
+耆
+耋
+而
+耍
+耐
+耒
+耕
+耗
+耘
+耙
+耦
+耨
+耳
+耶
+耷
+耸
+耻
+耽
+耿
+聂
+聆
+聊
+聋
+职
+聒
+联
+聘
+聚
+聪
+聴
+聿
+肃
+肄
+肆
+肇
+肉
+肋
+肌
+肏
+肓
+肖
+肘
+肚
+肛
+肝
+肠
+股
+肢
+肤
+肥
+肩
+肪
+肮
+肯
+肱
+育
+肴
+肺
+肽
+肾
+肿
+胀
+胁
+胃
+胄
+胆
+背
+胍
+胎
+胖
+胚
+胛
+胜
+胝
+胞
+胡
+胤
+胥
+胧
+胫
+胭
+胯
+胰
+胱
+胳
+胴
+胶
+胸
+胺
+能
+脂
+脆
+脇
+脉
+脊
+脍
+脏
+脐
+脑
+脓
+脖
+脘
+脚
+脣
+脩
+脯
+脱
+脲
+脳
+脸
+脾
+腆
+腈
+腊
+腋
+腌
+腐
+腑
+腓
+腔
+腕
+腥
+腩
+腭
+腮
+腰
+腱
+腴
+腹
+腺
+腻
+腼
+腾
+腿
+膀
+膈
+膊
+膏
+膑
+膘
+膛
+膜
+膝
+膦
+膨
+膳
+膺
+膻
+臀
+臂
+臃
+臆
+臊
+臓
+臣
+臧
+自
+臬
+臭
+至
+致
+臻
+臼
+臾
+舀
+舂
+舅
+舆
+舌
+舍
+舎
+舐
+舒
+舔
+舖
+舗
+舛
+舜
+舞
+舟
+航
+舫
+般
+舰
+舱
+舵
+舶
+舷
+舸
+船
+舺
+舾
+艇
+艋
+艘
+艮
+良
+艰
+色
+艳
+艹
+艺
+艾
+节
+芃
+芈
+芊
+芋
+芍
+芎
+芒
+芙
+芜
+芝
+芡
+芥
+芦
+芩
+芪
+芫
+芬
+芭
+芮
+芯
+花
+芳
+芷
+芸
+芹
+芽
+芾
+苁
+苄
+苇
+苋
+苍
+苏
+苑
+苒
+苓
+苔
+苕
+苗
+苛
+苜
+苞
+苟
+苡
+苣
+若
+苦
+苫
+苯
+英
+苷
+苹
+苻
+茁
+茂
+范
+茄
+茅
+茉
+茎
+茏
+茗
+茜
+茧
+茨
+茫
+茬
+茭
+茯
+茱
+茴
+茵
+茶
+茸
+茹
+茼
+荀
+荃
+荆
+草
+荏
+荐
+荒
+荔
+荖
+荘
+荚
+荞
+荟
+荠
+荡
+荣
+荤
+荥
+荧
+荨
+荪
+荫
+药
+荳
+荷
+荸
+荻
+荼
+荽
+莅
+莆
+莉
+莎
+莒
+莓
+莘
+莞
+莠
+莪
+莫
+莱
+莲
+莴
+获
+莹
+莺
+莽
+莿
+菀
+菁
+菅
+菇
+菈
+菊
+菌
+菏
+菓
+菖
+菘
+菜
+菟
+菠
+菡
+菩
+菱
+菲
+菸
+菽
+萁
+萃
+萄
+萋
+萌
+萍
+萎
+萘
+萝
+萤
+营
+萦
+萧
+萨
+萩
+萱
+萸
+萼
+落
+葆
+著
+葚
+葛
+葡
+董
+葩
+葫
+葬
+葭
+葱
+葳
+葵
+葺
+蒂
+蒋
+蒐
+蒙
+蒜
+蒟
+蒡
+蒨
+蒲
+蒸
+蒹
+蒻
+蒿
+蓁
+蓄
+蓆
+蓉
+蓑
+蓓
+蓖
+蓝
+蓟
+蓦
+蓬
+蓼
+蓿
+蔑
+蔓
+蔗
+蔘
+蔚
+蔡
+蔫
+蔬
+蔵
+蔷
+蔺
+蔻
+蔼
+蔽
+蕃
+蕈
+蕉
+蕊
+蕙
+蕤
+蕨
+蕲
+蕴
+蕻
+蕾
+薄
+薅
+薇
+薏
+薑
+薙
+薛
+薨
+薪
+薬
+薯
+薰
+薹
+藏
+藐
+藓
+藕
+藜
+藤
+藩
+藻
+藿
+蘑
+蘸
+蘼
+虎
+虏
+虐
+虑
+虔
+虚
+虞
+虢
+虫
+虬
+虱
+虹
+虻
+虽
+虾
+蚀
+蚁
+蚂
+蚊
+蚌
+蚓
+蚕
+蚜
+蚝
+蚣
+蚤
+蚩
+蚪
+蚯
+蚱
+蚵
+蛀
+蛆
+蛇
+蛊
+蛋
+蛎
+蛐
+蛔
+蛙
+蛛
+蛟
+蛤
+蛭
+蛮
+蛰
+蛳
+蛹
+蛾
+蜀
+蜂
+蜃
+蜇
+蜈
+蜊
+蜍
+蜒
+蜓
+蜕
+蜗
+蜘
+蜚
+蜜
+蜡
+蜢
+蜥
+蜱
+��
+蜷
+蜻
+蜿
+蝇
+蝈
+蝉
+蝌
+蝎
+蝗
+蝙
+蝠
+蝨
+蝴
+蝶
+蝼
+螂
+螃
+融
+螨
+螯
+螳
+螺
+蟀
+蟆
+蟋
+蟑
+蟒
+蟠
+蟹
+蟾
+蠍
+蠔
+蠕
+蠛
+蠡
+蠢
+蠹
+血
+衄
+衅
+行
+衍
+衔
+街
+衙
+衞
+衡
+衢
+衣
+补
+表
+衩
+衫
+衬
+衮
+衰
+衲
+衷
+衾
+衿
+袁
+袂
+袄
+袅
+袈
+袋
+袍
+袒
+袖
+袜
+袤
+袪
+被
+袭
+袱
+裁
+裂
+装
+裆
+裔
+裕
+裘
+裙
+裟
+裤
+裨
+裱
+裳
+裴
+裸
+裹
+裾
+褂
+褐
+褒
+褓
+褔
+褚
+褥
+褪
+褫
+褶
+襁
+襄
+襟
+西
+要
+覃
+覆
+覇
+覚
+覧
+観
+见
+观
+规
+觅
+视
+览
+觉
+觊
+觎
+觐
+觑
+角
+觞
+解
+觥
+触
+言
+訳
+証
+詹
+誉
+誓
+読
+諡
+譁
+警
+譬
+譲
+讚
+计
+订
+认
+讥
+讧
+讨
+让
+讪
+讫
+训
+议
+讯
+记
+讲
+讳
+讴
+讶
+讷
+许
+讹
+论
+讼
+讽
+设
+访
+诀
+证
+诃
+评
+诅
+识
+诈
+诉
+诊
+诋
+词
+诏
+译
+试
+诗
+诘
+诙
+诚
+诛
+话
+诞
+诟
+诠
+诡
+询
+诣
+诤
+该
+详
+诧
+诩
+诫
+诬
+语
+误
+诰
+诱
+诲
+说
+诵
+诶
+请
+诸
+诺
+读
+诽
+课
+诿
+谀
+谁
+调
+谄
+谅
+谆
+谈
+谊
+谋
+谌
+谍
+谎
+谏
+谐
+谑
+谒
+谓
+谔
+谕
+谗
+谘
+谙
+谚
+谛
+谜
+谟
+谢
+谣
+谤
+谥
+谦
+谧
+谨
+谩
+谪
+谬
+谭
+谯
+谱
+谲
+谴
+谶
+谷
+豁
+豆
+豇
+豉
+豊
+豌
+豔
+豚
+象
+豢
+豪
+豫
+豹
+豺
+貂
+貅
+貌
+貔
+貘
+贝
+贞
+负
+贡
+财
+责
+贤
+败
+账
+货
+质
+贩
+贪
+贫
+贬
+购
+贮
+贯
+贰
+贱
+贲
+贴
+贵
+贷
+贸
+费
+贺
+贻
+贼
+贾
+贿
+赁
+赂
+赃
+资
+赅
+赈
+赊
+赋
+赌
+赎
+赏
+赐
+赓
+赔
+赖
+赘
+赚
+赛
+赝
+赞
+赠
+赡
+赢
+赣
+赤
+赦
+赧
+赫
+赭
+走
+赳
+赴
+赵
+赶
+起
+趁
+超
+越
+趋
+趟
+趣
+足
+趴
+趵
+趸
+趺
+趾
+跃
+跄
+跆
+跋
+跌
+跎
+跑
+跖
+跚
+跛
+距
+跟
+跤
+跨
+跩
+跪
+路
+跳
+践
+跷
+跹
+跺
+跻
+踉
+踊
+踌
+踏
+踝
+踞
+踟
+踢
+踩
+踪
+踮
+踱
+踵
+踹
+蹂
+蹄
+蹇
+蹈
+蹉
+蹊
+蹋
+蹑
+蹒
+蹙
+蹟
+蹦
+蹩
+蹬
+蹭
+蹲
+蹴
+蹶
+蹼
+蹿
+躁
+躇
+躏
+身
+躬
+躯
+躲
+躺
+転
+軽
+车
+轧
+轨
+轩
+转
+轭
+轮
+软
+轰
+轲
+轴
+轶
+轻
+轼
+载
+轿
+较
+辄
+辅
+辆
+辇
+辈
+辉
+辊
+辍
+辐
+辑
+输
+辕
+辖
+辗
+辘
+辙
+辛
+辜
+辞
+辟
+辣
+辨
+辩
+辫
+辰
+辱
+边
+辺
+辻
+込
+辽
+达
+迁
+迂
+迄
+迅
+过
+迈
+迎
+运
+近
+返
+还
+这
+进
+远
+违
+连
+迟
+迢
+迤
+迥
+迦
+迩
+迪
+迫
+迭
+述
+迷
+迸
+迹
+迺
+追
+退
+送
+适
+逃
+逅
+逆
+选
+逊
+逍
+透
+逐
+递
+途
+逗
+通
+逛
+逝
+逞
+速
+造
+逢
+逮
+逵
+逶
+逸
+逻
+逼
+逾
+遁
+遂
+遅
+遇
+遍
+遏
+遐
+遑
+遒
+道
+遗
+遛
+遢
+遣
+遥
+遨
+遭
+遮
+遴
+遵
+遶
+遽
+避
+邀
+邂
+邃
+邈
+邋
+邑
+邓
+邕
+邛
+邝
+邢
+那
+邦
+邨
+邪
+邬
+邮
+邯
+邰
+邱
+邳
+邵
+邸
+邹
+邺
+邻
+郁
+郅
+郊
+郎
+郑
+郜
+郝
+郡
+郢
+郤
+郦
+郧
+部
+郫
+郭
+郴
+郷
+郸
+都
+鄂
+鄙
+鄞
+鄢
+鄱
+酉
+酊
+酋
+酌
+配
+酐
+酒
+酗
+酚
+酝
+酢
+酣
+酥
+酩
+酪
+酬
+酮
+酯
+酰
+酱
+酵
+酶
+酷
+酸
+酿
+醃
+醇
+醉
+醋
+醍
+醐
+醒
+醚
+醛
+醣
+醪
+醮
+醯
+醴
+醺
+采
+釉
+释
+里
+重
+野
+量
+金
+釜
+釦
+鈪
+鉄
+鉴
+銭
+銮
+鍊
+鎌
+鎏
+鎗
+鏖
+鑑
+鑫
+针
+钉
+钊
+钎
+钏
+钒
+钓
+钗
+钙
+钛
+钜
+钝
+钞
+钟
+钠
+钡
+钢
+钣
+钤
+钥
+钦
+钧
+钨
+钩
+钮
+钯
+钰
+钱
+钳
+钴
+钵
+钺
+钻
+钼
+钾
+钿
+铀
+铁
+铂
+铃
+铄
+铅
+铆
+铉
+铎
+铐
+铛
+铜
+铝
+铠
+铡
+铢
+铣
+铤
+铨
+铩
+铬
+铭
+铮
+铰
+铲
+铵
+银
+铸
+铺
+链
+铿
+销
+锁
+锂
+锄
+锅
+锆
+锈
+锉
+锋
+锌
+锏
+锐
+锑
+错
+锚
+锟
+锡
+锢
+锣
+锤
+锥
+锦
+锭
+键
+锯
+锰
+锲
+锵
+锹
+锺
+锻
+镀
+镁
+镂
+镇
+镉
+镌
+镍
+镐
+镑
+镕
+镖
+镗
+镛
+镜
+镣
+镭
+镯
+镰
+镳
+镶
+长
+閒
+関
+闇
+闘
+闢
+门
+闪
+闫
+闭
+问
+闯
+闰
+闲
+间
+闵
+闷
+闸
+闹
+闺
+闻
+闽
+闾
+阀
+阁
+阂
+阅
+阆
+阇
+阈
+阉
+阎
+阐
+阑
+阔
+阕
+阖
+阙
+阚
+阜
+队
+阡
+阮
+阱
+防
+阳
+阴
+阵
+阶
+阻
+阿
+陀
+陂
+附
+际
+陆
+陇
+陈
+陋
+陌
+降
+限
+陕
+陛
+陞
+陟
+陡
+院
+除
+陨
+险
+陪
+陲
+陵
+陶
+陷
+険
+隅
+隆
+隈
+隋
+隍
+随
+隐
+隔
+隘
+隙
+障
+隠
+隣
+隧
+隶
+隼
+隽
+难
+雀
+雁
+雄
+雅
+集
+雇
+雉
+雌
+雍
+雎
+雏
+雑
+雒
+雕
+雨
+雪
+雯
+雰
+雳
+零
+雷
+雹
+雾
+需
+霁
+霄
+霆
+震
+霈
+霉
+霊
+霍
+霎
+霏
+霑
+霓
+霖
+霜
+霞
+霭
+霰
+露
+霸
+霹
+霾
+青
+靓
+靖
+静
+靛
+非
+靠
+靡
+面
+靥
+革
+靳
+靴
+靶
+靼
+鞅
+鞋
+鞍
+鞑
+鞘
+鞠
+鞣
+鞭
+韦
+韧
+韩
+韬
+韭
+音
+韵
+韶
+頫
+頼
+页
+顶
+顷
+项
+顺
+须
+顼
+顽
+顾
+顿
+颁
+颂
+预
+颅
+领
+颇
+颈
+颉
+颊
+颌
+颍
+颐
+频
+颓
+颔
+颖
+颗
+题
+颚
+颛
+颜
+额
+颞
+颠
+颡
+颢
+颤
+颦
+颧
+风
+飒
+飓
+飕
+飘
+飙
+飚
+飞
+食
+飨
+餐
+餮
+餵
+饍
+饕
+饥
+饨
+饪
+饬
+饭
+饮
+饯
+饰
+饱
+饲
+饴
+饵
+饶
+��
+饺
+饼
+饽
+饿
+馀
+馁
+馄
+馅
+馆
+馈
+馋
+馍
+馏
+馒
+馔
+首
+馗
+香
+馥
+馨
+駄
+駅
+駆
+験
+騨
+驒
+马
+驭
+驮
+驯
+驰
+驱
+驳
+驴
+驶
+驷
+驸
+驹
+驻
+驼
+驾
+驿
+骁
+骂
+骄
+骅
+骆
+骇
+骈
+骊
+骋
+验
+骏
+骐
+骑
+骗
+骚
+骛
+骜
+骞
+骠
+骡
+骤
+骥
+骧
+骨
+骰
+骶
+骷
+骸
+骼
+髂
+髅
+髋
+髓
+高
+髦
+髪
+髯
+髻
+鬃
+鬓
+鬟
+鬣
+鬼
+魁
+魂
+魄
+魅
+魇
+魍
+魏
+魔
+鮨
+鱼
+鱿
+鲁
+鲈
+鲍
+鲑
+鲛
+鲜
+鲟
+鲢
+鲤
+鲨
+鲫
+鲱
+鲲
+鲶
+鲷
+鲸
+鳃
+鳄
+鳅
+鳌
+鳍
+鳕
+鳖
+鳗
+鳝
+鳞
+鵰
+鸟
+鸠
+鸡
+鸢
+鸣
+鸥
+鸦
+鸨
+鸪
+鸭
+鸯
+鸳
+鸵
+鸽
+鸾
+鸿
+鹂
+鹃
+鹄
+鹅
+鹈
+鹉
+鹊
+鹌
+鹏
+鹑
+鹕
+鹘
+鹜
+鹞
+鹤
+鹦
+鹧
+鹫
+鹭
+鹰
+鹳
+鹿
+麂
+麋
+麒
+麓
+麝
+麟
+麦
+麴
+麸
+麺
+麻
+麾
+黄
+黍
+黎
+黏
+黑
+黒
+黔
+默
+黛
+黜
+黝
+黠
+黯
+鼋
+鼎
+鼐
+鼓
+鼠
+鼬
+鼹
+鼻
+鼾
+齁
+齐
+齢
+齿
+龄
+龅
+龈
+龊
+龋
+龌
+龙
+龚
+龛
+龟
+呡
+乾
+绗
+楦
+硌
+袢
+钕
+蕞
+癀
+皲
+貉
+唛
+笕
+椴
+―
+胗
+旯
+鳙
+鲇
+鳐
+鳜
+鲅
+鳊
+鲳
+鲽
+鲣
+枞
+炝
+醅
+馊
+捯
+抻
+绉
+馐
+饧
+莜
+嘬
+腘
+肫
+鳟
+镊
+犽
+洌
+蝰
+铱
+髌
+锃
+镲
+锗
+甑
+戗
+裥
+弎
+粝
+霂
+猄
+轱
+苎
+偲
+兿
+铷
+栢
+帏
+黢
+洇
+沄
+誊
+忸
+怩
+蚬
+籺
+氚
+犇
+锒
+鸩
+噘
+偾
+髫
+︰
+︱
+︶
+︿
+﹁
+﹂
+﹍
+﹏
+﹐
+﹑
+﹒
+﹔
+﹕
+﹖
+﹗
+﹙
+﹚
+﹝
+﹞
+﹡
+﹣
+！
+（
+）
+，
+：
+；
+？
+｜
+～
+｡
+｢
+｣
+､
+･
+ｯ
+ｰ
+ﾞ
+ﾟ
+￣
+￥
+...
+yam
+lofter
+##s
+by
+##0
+com
+##a
+##2
+##1
+##3
+##e
+##8
+##5
+##6
+##4
+##9
+##7
+##t
+##o
+##d
+##i
+##n
+app
+www
+the
+##m
+##c
+##l
+##y
+##r
+##g
+http
+qq
+##p
+##f
+google
+pixnet
+cookies
+tripadvisor
+##er
+##k
+##h
+facebook
+##b
+of
+##x
+##u
+iphone
+ip
+in
+##w
+##ing
+ctrip
+##on
+##v
+to
+id
+it
+windows
+llc
+top
+led
+at
+##an
+##z
+android
+and
+vr
+blogthis
+twitter
+##le
+ok
+cn
+no
+ios
+##in
+##mm
+on
+te
+ig
+lv
+##ng
+##us
+pc
+──
+##te
+##ed
+html
+ncc
+wifi
+email
+blog
+is
+mail
+online
+##al
+dvd
+##ic
+studio
+##℃
+##ia
+line
+vip
+##q
+##ce
+##en
+for
+##is
+##ra
+##es
+##j
+usb
+net
+cp
+asia
+##cm
+diy
+new
+ta
+language
+vs
+apple
+tw
+web
+##ne
+ipad
+you
+##re
+##tion
+ps
+de
+bt
+pony
+atm
+##ch
+ceo
+##or
+go
+##na
+av
+pro
+cafe
+pinterest
+pixstyleme3c
+##ta
+more
+said
+mp3
+##ll
+nba
+jun
+tv
+pm
+nbsp
+##ie
+linux
+##ma
+cd
+hd
+##ion
+am
+##th
+##st
+##se
+##et
+gdp
+my
+abc
+flash
+one
+##ck
+gps
+##ly
+web885
+##ge
+xd
+boss
+isbn
+org
+##ry
+me
+love
+##ter
+##ar
+##la
+hotel
+pk
+ie
+##os
+##el
+seo
+cpu
+##ml
+p2p
+may
+sun
+tue
+internet
+cc
+posted
+youtube
+##at
+##man
+ii
+abs
+nt
+pdf
+yahoo
+ago
+##it
+news
+mac
+##me
+java
+spa
+##de
+##nt
+hk
+all
+plus
+la
+##mb
+##ve
+west
+##da
+air
+##ps
+##to
+logo
+htc
+php
+https
+fi
+momo
+##son
+sat
+##ke
+ebd
+suv
+wi
+day
+apk
+##um
+mv
+galaxy
+wiki
+or
+brake
+this
+mon
+po
+javascript
+life
+home
+june
+##ss
+system
+pp
+world
+fb
+br
+##as
+ic
+ai
+leonardo
+safari
+live
+free
+xx
+wed
+win7
+kiehl
+##co
+lg
+o2o
+##go
+us
+mm
+vfm
+kanye
+##id
+jr
+##ey
+rss
+##sa
+##ro
+##am
+##no
+thu
+fri
+##sh
+##ki
+comments
+name
+##pe
+##ine
+max
+uber
+##mi
+##ton
+wordpress
+office
+##ment
+bd
+win10
+##ld
+##li
+gmail
+bb
+dior
+##rs
+##ri
+##rd
+up
+cad
+dr
+read
+##io
+url
+pvc
+paypal
+show
+policy
+##ty
+with
+txt
+##ba
+dna
+from
+post
+mini
+ar
+taiwan
+john
+##ga
+privacy
+agoda
+##ny
+word
+##by
+##ur
+##hz
+##ang
+cookie
+netscape
+##ka
+##～
+##ad
+house
+share
+note
+ibm
+code
+hello
+nike
+sim
+survey
+wikia
+cbc
+##tor
+##kg
+##rt
+campaign
+store
+os
+##ct
+##ts
+##°
+api
+##ns
+excel
+##ao
+##nd
+university
+##ya
+##il
+pierre
+ipo
+hotels
+##ian
+years
+##ers
+high
+##day
+time
+##ay
+bug
+##line
+##be
+xp
+talk2yam
+yamservice
+coco
+##dy
+sony
+##ies
+microsoft
+david
+people
+##ha
+instagram
+intel
+##ot
+iso
+##va
+##mo
+##land
+xxx
+man
+co
+ltxsw
+##ation
+baby
+##pa
+##ol
+tag
+##ue
+msn
+oppo
+##ca
+control
+##om
+st
+chrome
+##ure
+be
+lol
+##bo
+lady
+##way
+##ko
+##do
+##un
+corporation
+##ni
+herme
+##up
+ui
+##ds
+ppt
+admin
+three
+bbc
+re
+ca
+hp
+##ee
+tpp
+##ive
+root
+##cc
+##ble
+##ity
+adobe
+park
+et
+oled
+city
+##ex
+##ler
+##ap
+china
+##book
+view
+##ice
+global
+##km
+your
+hong
+##mg
+out
+##ms
+ng
+ebay
+menu
+ubuntu
+##cy
+rom
+##view
+open
+ktv
+do
+server
+##lo
+if
+english
+##oo
+step1
+kong
+club
+july
+inc
+mr
+hi
+##net
+touch
+##ls
+##ii
+michael
+lcd
+phone
+james
+step2
+ios9
+##box
+dc
+##ley
+samsung
+pokemon
+css
+##ent
+##les
+s8
+atom
+play
+bmw
+##said
+sa
+etf
+ctrl
+adidas
+amazon
+##ber
+##ner
+visa
+##der
+connectivity
+##hi
+firefox
+hr
+so
+style
+mark
+pop
+ol
+skip
+as
+##ir
+mba
+##ai
+le
+##ver
+cafe2017
+lte
+super
+##ron
+amd
+like
+are
+##ster
+we
+##sk
+paul
+data
+international
+##ft
+longchamp
+ssd
+good
+##ti
+reply
+##my
+apr
+star
+##ker
+source
+js
+get
+force
+photo
+##one
+##ow
+link
+bbs
+goods
+##lin
+python
+##ip
+game
+##ics
+blue
+page
+itunes
+gt
+gif
+##ff
+group
+about
+bar
+ganji
+##nce
+music
+lee
+not
+##per
+an
+faq
+comment
+days
+##ock
+##bs
+v1
+player
+xbox
+sql
+fm
+f1
+##ah
+##lv
+##mp
+melody
+xml
+market
+##au
+what
+gl
+##age
+tips
+book
+##ting
+mysql
+can
+##ung
+wonderland
+watch
+##ction
+mar
+mobile
+article
+##db
+part
+party
+##ore
+##op
+dj
+main
+##ong
+art
+ad
+pm2
+japan
+ts
+##ica
+der
+sm
+##wa
+ct
+homemesh
+search
+##tv
+##di
+macbook
+service
+type
+##ier
+##si
+##ok
+best
+goris
+lock
+cf
+big
+##ut
+ftp
+carol
+##vi
+happy
+sd
+##ac
+anti
+pe
+cnn
+iii
+esp
+jan
+tags
+august
+vol
+##fs
+##sion
+design
+ac
+press
+jordan
+ppp
+that
+key
+check
+##tt
+##㎡
+##lt
+power
+##bc
+vivi
+he
+jpg
+##rry
+nb
+##ted
+##rn
+usd
+##t00
+master
+model
+al
+ram
+goo
+##ui
+red
+##ary
+rpg
+item
+##pm
+##za
+project
+hot
+td
+blogabstract
+##ger
+gr2
+black
+electronic
+nfc
+year
+asus
+html5
+cindy
+##hd
+m3
+esc
+##od
+booking
+fed
+tvb
+##ina
+mit
+chan
+distribution
+next
+peter
+bios
+steam
+cm
+pk10
+##ix
+dec
+nasa
+##ana
+icecat
+b1
+will
+li
+se
+##ji
+##ard
+oct
+##ain
+jp
+##ze
+##bi
+cio
+smart
+h5
+##port
+curve
+vpn
+##nm
+##dia
+utc
+rmvb
+chanel
+a4
+miss
+##and
+##im
+media
+who
+she
+girl
+vera
+class
+vivo
+king
+##ei
+national
+ab
+ipod
+ap
+ms
+mp4
+msci
+##po
+mg
+index
+##bit
+##out
+##zz
+apec
+photoshop
+opec
+##tes
+##ast
+○○
+##ling
+##ory
+##ical
+kitty
+content
+step3
+##cn
+win8
+vc
+iphone7
+robert
+tcl
+beauty
+en
+dollars
+##ys
+##oc
+step
+pay
+yy
+a1
+##lly
+##ks
+download
+sep
+exe
+ph
+school
+gb
+center
+pr
+street
+##board
+uv
+##lan
+winrar
+##que
+##ua
+##com
+gpu
+ettoday
+fu
+tom
+##ren
+##via
+b2b
+##tch
+rose
+arm
+mb
+##ial
+##nn
+nvidia
+step4
+mvp
+york
+how
+cpi
+gov
+kg
+joe
+##xx
+mandy
+pa
+##ser
+copyright
+fashion
+don
+ecu
+##ist
+##art
+erp
+wap
+have
+##lm
+talk
+##ek
+##ning
+##if
+ch
+##ite
+video
+cs
+san
+iot
+look
+##ku
+october
+##ux
+trump
+##hs
+##ide
+box
+first
+##ins
+april
+##ight
+angel
+protected
+aa
+x1
+m2
+##fe
+##×
+##ho
+size
+min
+ofo
+fun
+gomaji
+ex
+hdmi
+food
+dns
+march
+chris
+kevin
+##lla
+##pp
+##ec
+ag
+ems
+##rm
+##ham
+off
+asp
+team
+fandom
+ed
+##ell
+info
+sina
+##able
+##ctor
+dll
+rights
+ltd
+idc
+jul
+ma
+surface
+mall
+eps
+green
+map
+space
+donald
+v2
+sodu
+##light
+reserved
+htm
+##han
+mod
+##ise
+##tions
+ti
+##shi
+doc
+icp
+wang
+##ram
+shopping
+aug
+##pi
+##well
+now
+wam
+b2
+##hu
+##gb
+f2
+mix
+##ef
+##uan
+bwl
+##plus
+##res
+core
+##ess
+tea
+hktvmall
+nhk
+##ate
+list
+##ese
+feb
+inn
+nov
+daniel
+##ci
+pass
+##bet
+##nk
+coffee
+ssl
+airbnb
+##ute
+fbi
+woshipm
+skype
+ea
+cg
+sp
+##fc
+##www
+yes
+edge
+alt
+fpga
+##ght
+##gs
+iso9001
+##ile
+##wood
+##uo
+image
+lin
+icon
+american
+##em
+set
+says
+##king
+##tive
+blogger
+##ox
+##zy
+##red
+##ium
+##lf
+nokia
+claire
+##ding
+november
+lohas
+##tic
+##cs
+##che
+##ire
+##gy
+##ult
+db
+january
+win
+road
+ptt
+##fa
+##mer
+anna
+pchome
+udn
+ef
+##time
+##tte
+g20
+white
+garden
+eleven
+di
+chen
+young
+cosplay
+bat
+##tra
+kindle
+npc
+steve
+etc
+##ern
+call
+xperia
+ces
+travel
+sk
+s7
+##ous
+##int
+edu
+file
+cho
+qr
+##car
+##our
+##ant
+eric
+rends
+##jo
+mastercard
+kb
+##min
+##ino
+vista
+##ris
+##ud
+jack
+##set
+pos
+##her
+##ou
+taipei
+beta
+##fi
+express
+body
+##ill
+aphojoy
+user
+december
+meiki
+##ick
+tweet
+richard
+##av
+iphone6
+##dd
+views
+##mark
+pd
+times
+level
+##ash
+point
+##ome
+koreanmall
+##ak
+george
+q2
+wma
+tcp
+full
+mlb
+##lle
+##watch
+tm
+run
+smith
+business
+##und
+color
+##tal
+##less
+moon
+##rl
+update
+pcb
+shop
+little
+end
+##mhz
+van
+dsp
+easy
+##house
+##key
+history
+oh
+##hy
+##web
+oem
+let
+was
+##gg
+review
+##wan
+##°c
+uc
+title
+##val
+united
+##ons
+doi
+trivago
+overdope
+sbs
+##ance
+grand
+special
+imf
+wx17house
+##so
+audi
+##he
+london
+william
+##rp
+##ake
+science
+beach
+cfa
+amp
+ps4
+##link
+##hp
+crm
+ferragamo
+bell
+make
+##eng
+under
+zh
+photos
+##style
+via
+da
+##gi
+company
+i7
+##ray
+thomas
+ufo
+i5
+##max
+plc
+ben
+back
+research
+mike
+##pc
+september
+##ace
+vps
+february
+pantos
+wp
+lisa
+jquery
+night
+long
+offer
+##berg
+##news
+ray
+fks
+wto
+over
+##all
+##rus
+##works
+blogtitle
+loftpermalink
+martin
+test
+ling
+km
+fda
+v3
+##ja
+outlet
+family
+##ea
+##top
+story
+##ness
+salvatore
+##lu
+swift
+room
+oracle
+##ul
+sam
+b2c
+week
+pi
+rock
+##ean
+##gle
+cctv
+after
+chinese
+##back
+powered
+x2
+##tan
+##nes
+canon
+only
+##zi
+##las
+say
+##oe
+##sd
+##bot
+##world
+##zo
+sky
+made
+top100
+just
+pmi
+gap
+##vr
+les
+ball
+vogue
+vi
+ing
+ofweek
+cos
+##list
+##ort
+##lon
+last
+##tc
+##of
+##bus
+##gen
+real
+eva
+a3
+nas
+##lie
+##ria
+##coin
+##bt
+his
+cat
+nata
+vive
+health
+drive
+sir
+du
+cup
+##ook
+##sy
+alex
+msg
+tour
+##word
+ebooks
+r8
+block
+nice
+pvp
+months
+rewards
+##ther
+##xi
+##sc
+micro
+gg
+blogfp
+op
+daily
+m1
+true
+##bb
+ml
+##tar
+##ky
+anthony
+##yo
+state
+##ara
+##aa
+##rc
+##tz
+##ston
+gear
+##eo
+##ade
+ge
+see
+##win
+##ura
+ss
+heart
+##den
+##ita
+down
+##sm
+el
+png
+rakuten
+whatsapp
+bay
+dream
+add
+##use
+pad
+gucci
+mpv
+##ode
+##fo
+island
+jason
+chicago
+##hone
+io
+sogo
+be2
+##ology
+cloud
+vcd
+##con
+##ford
+##joy
+##kb
+##rade
+but
+##ach
+docker
+##ful
+rfid
+ul
+##ase
+hit
+ford
+##star
+a2
+sdk
+reading
+edited
+##are
+cmos
+##mc
+siri
+light
+##ella
+bloomberg
+##read
+pizza
+##ison
+jimmy
+##vm
+college
+node
+journal
+ba
+##play
+##cer
+magic
+##yu
+jump
+tt
+##ings
+asr
+##lia
+step5
+network
+##cd
+mc
+pixstyleme
+money
+bl
+act
+##tus
+tokyo
+##rial
+##life
+emba
+##ae
+saas
+tcs
+##rk
+##wang
+summer
+##sp
+ko
+##ving
+premium
+netflix
+uk
+mt
+##lton
+right
+frank
+two
+##ple
+##cal
+##sen
+##ville
+hold
+nexus
+dd
+##ius
+##mah
+tila
+zero
+ce
+##tin
+resort
+##ws
+charles
+old
+p10
+report
+##ru
+bus
+vans
+lt
+##est
+pv
+links
+rebecca
+##dm
+azure
+limited
+bit
+##mon
+moto
+##eam
+var
+eos
+blogspot
+e3
+dos
+dm
+fc
+##ments
+##ik
+##kw
+boy
+##bin
+##ata
+er
+##vin
+##tu
+##ula
+station
+##ature
+files
+zara
+hdr
+top10
+nature
+magazine
+s6
+marriott
+avira
+case
+tab
+##ran
+tony
+##home
+oculus
+im
+##ral
+jean
+saint
+cry
+rosie
+##force
+##ini
+ice
+##bert
+##nder
+##mber
+pet
+plurk
+##sis
+##ence
+tim
+##nc
+##name
+log
+ips
+great
+ikea
+malaysia
+unix
+##ncy
+##nie
+akb48
+##ye
+##oid
+##chi
+oa
+xuehai
+##orm
+##rf
+##ware
+ho
+##pro
+text
+##era
+bob
+##ub
+scp
+avi
+##zen
+mi
+wu
+museum
+qvod
+apache
+lake
+jcb
+ni
+##hr
+hill
+ne
+weibo
+ruby
+##row
+iv
+##ish
+github
+mate
+##lot
+##ane
+andrew
+##tina
+t1
+rf
+ed2k
+##vel
+way
+final
+ns
+sweet
+bytes
+##ene
+##cker
+##px
+topapp
+helpapp
+rs
+low
+g4g
+care
+ldquo
+##fork
+leave
+rm
+edition
+##gan
+##zon
+##qq
+##google
+##ism
+gold
+explorer
+##zer
+toyota
+category
+select
+visual
+##labels
+restaurant
+##md
+posts
+s1
+##ico
+angelababy
+sports
+s3
+mbc
+shell
+x86
+candy
+##new
+kbs
+face
+xl
+##here
+swissinfo
+v8
+dram
+##ual
+##vice
+##wer
+sport
+q1
+ios10
+public
+int
+card
+ep
+au
+rt
+bill
+##mll
+kim
+wan
+##uk
+x3
+scott
+##ming
+e5
+h7n9
+worldcat
+brown
+##vo
+##led
+##ax
+##ert
+paris
+polo
+##lr
+capital
+##hing
+bank
+cv
+##chat
+adc
+##ule
+digital
+hotmail
+##pad
+bbq
+quot
+##ring
+before
+wali
+mcu
+costco
+north
+switch
+##city
+philips
+##mann
+management
+panasonic
+##cl
+##vd
+##ping
+##rge
+alice
+##lk
+css3
+##ney
+vision
+alpha
+##ular
+##tter
+lz
+mode
+gre
+pci
+##tm
+##yan
+##let
+work
+war
+coach
+ah
+mary
+huang
+##pt
+a8
+pt
+follow
+##berry
+##ew
+a5
+ghost
+##wn
+##og
+south
+##code
+girls
+##rid
+action
+villa
+git
+r11
+table
+games
+##cket
+error
+##anonymoussaid
+##ag
+here
+##ame
+##gc
+qa
+##lis
+gmp
+##gin
+vmalife
+##cher
+yu
+wedding
+##tis
+demo
+dragon
+soho
+social
+bye
+##rant
+river
+orz
+acer
+##ats
+del
+##ven
+ups
+value
+macd
+yougou
+##dn
+##ano
+ll
+##urt
+##rent
+continue
+script
+##wen
+##ect
+paper
+shift
+##chel
+##cat
+x5
+fox
+car
+aaa
+##blog
+loading
+##yn
+##tp
+kuso
+si
+sns
+rmb
+vdc
+forest
+central
+prime
+help
+ultra
+##rmb
+square
+##field
+##reen
+##ors
+##ju
+c1
+start
+##air
+##map
+cdn
+##wo
+cba
+stephen
+m8
+##get
+opera
+##base
+##ood
+vsa
+com™
+##aw
+##ail
+count
+t2
+##een
+hop
+##gp
+vsc
+tree
+##eg
+##ose
+##ories
+##shop
+alphago
+v4
+simon
+fluke62max
+zip
+##sta
+louis
+cr
+bas
+bc
+##yer
+hadoop
+##ube
+##wi
+hola
+##low
+place
+centre
+d3
+##fer
+##media
+exchange
+series
+##san
+eb
+##bank
+q3
+##nge
+##mail
+take
+##lp
+client
+east
+cache
+event
+vincent
+##nse
+sui
+adchoice
+##stry
+##zone
+ga
+apps
+sea
+##ab
+cisco
+##rner
+kymco
+##care
+dha
+##pu
+##yi
+minkoff
+royal
+p1
+annie
+collection
+kpi
+playstation
+bh
+##bar
+queen
+radio
+andy
+armani
+##xy
+manager
+iherb
+##ery
+##share
+spring
+raid
+johnson
+##ob
+volvo
+hall
+##ball
+v6
+our
+taylor
+##hk
+bi
+##cp
+kate
+bo
+water
+technology
+##rie
+##ona
+##sl
+hpv
+gtx
+hip
+rdquo
+jayz
+stone
+##lex
+##rum
+namespace
+##ale
+##atic
+des
+##erson
+##ql
+##ves
+##type
+enter
+d2
+##mix
+##bian
+a9
+jj
+ky
+##lc
+access
+movie
+##hc
+tower
+##ration
+##mit
+##nch
+ua
+tel
+prefix
+##o2
+##point
+ott
+##http
+##ury
+baidu
+##ink
+member
+##logy
+bigbang
+nownews
+##js
+##shot
+##tb
+eba
+##tics
+##lus
+v5
+spark
+##ama
+there
+##ions
+god
+##lls
+##down
+hiv
+##ress
+burberry
+day2
+##kv
+jeff
+related
+film
+edit
+joseph
+##ark
+cx
+order
+g9
+##ans
+##tty
+s5
+##bee
+thread
+xr
+buy
+sh
+land
+spotify
+mx
+##ari
+##verse
+sf
+why
+nego
+sunny
+dom
+exo
+positioning
+fit
+rgb
+##tton
+kiss
+alexa
+adam
+lp
+mp
+##ties
+##llow
+amy
+##du
+np
+institute
+##rth
+##lar
+##des
+sidebar
+imax
+site
+##cky
+##kit
+##ime
+season
+##fun
+gogoro
+a7
+pu
+lily
+fire
+twd600
+##vis
+##cture
+information
+close
+friday
+yi
+nick
+##tta
+##tel
+##lock
+cbd
+economy
+tinker
+double
+voice
+##app
+oops
+channel
+today
+##right
+raw
+xyz
+jim
+edm
+##cent
+supreme
+ds
+##its
+##asia
+dropbox
+##tti
+books
+##tle
+##ller
+##ken
+##more
+##boy
+sex
+##dom
+t3
+##ider
+##unch
+feel
+##put
+s2
+mo
+##gh
+men
+ka
+amoled
+div
+##tr
+##n1
+port
+howard
+##tags
+ken
+dnf
+##nus
+adsense
+ide
+buff
+thunder
+##town
+##ique
+has
+##body
+auto
+pin
+##erry
+tee
+number
+##the
+object
+psp
+cool
+udnbkk
+##mic
+miui
+##tro
+most
+r2
+##alk
+##nity
+s4
+law
+version
+##oa
+n1
+sgs
+docomo
+##tf
+##ack
+henry
+fc2
+##ded
+##sco
+##rite
+linkedin
+##ada
+##now
+wii
+##ndy
+ucbug
+sputniknews
+legalminer
+##ika
+##xp
+##bu
+q10
+oo
+b6
+come
+##rman
+cheese
+ming
+maker
+##gm
+nikon
+##fig
+ppi
+kelly
+jchere
+ted
+md
+fgo
+tech
+##tto
+dan
+soc
+##gl
+##len
+hair
+earth
+img
+##pper
+##a1
+acca
+##ition
+##ference
+suite
+##ig
+outlook
+##mond
+##cation
+##pr
+airport
+##over
+jones
+##ith
+lab
+##su
+co2
+town
+piece
+##llo
+no1
+vmware
+##qi
+focus
+reader
+##admin
+##ora
+tb
+false
+##log
+know
+lan
+##ces
+f4
+##ume
+motel
+stop
+##oper
+na
+flickr
+netcomponents
+##af
+pose
+williams
+local
+##ound
+##cg
+##site
+##iko
+gsm
+con
+##ath
+friends
+##hip
+cell
+##rey
+cream
+##cks
+##dp
+facebooktwitterpinterestgoogle
+sso
+shtml
+song
+swiss
+##mw
+lumia
+xdd
+string
+tiffany
+marc
+insee
+russell
+sc
+dell
+##ations
+camera
+##vs
+##flow
+##late
+classic
+##nter
+stay
+g1
+mtv
+##ever
+##lab
+##nger
+qe
+sata
+ryan
+d1
+cms
+##cing
+su
+editor
+##nap
+security
+sunday
+association
+##ens
+##bra
+acg
+sofascore
+mkv
+##ign
+jonathan
+gary
+build
+labels
+##oto
+tesla
+moba
+qi
+gohappy
+general
+ajax
+society
+##test
+##urs
+wps
+fedora
+##ich
+mozilla
+##dr
+usa
+urn
+##lina
+grace
+##die
+##try
+##ader
+elle
+##chen
+price
+##ten
+uhz
+##ough
+eq
+##hen
+states
+push
+session
+balance
+wow
+##cus
+##py
+when
+##ward
+##ep
+wong
+library
+prada
+##cle
+running
+##ree
+ck
+date
+q4
+##ctive
+##ool
+mk
+##ira
+die
+secret
+rq
+dota
+buffet
+e6
+##ez
+pan
+ha
+##card
+##cha
+alan
+day3
+eye
+f3
+##end
+france
+keep
+adi
+rna
+tvbs
+##ala
+solo
+nova
+##tail
+support
+##ries
+##ved
+base
+copy
+iis
+fps
+##ways
+hero
+hgih
+profile
+fish
+mu
+ssh
+entertainment
+chang
+##wd
+click
+cake
+##ond
+pre
+##tom
+kic
+pixel
+##ov
+##fl
+product
+##pd
+dear
+##gate
+es
+yumi
+audio
+##²
+##sky
+echo
+bin
+where
+##ture
+##ape
+find
+sap
+isis
+nand
+##load
+##ream
+band
+a6
+never
+##post
+festival
+##we
+guide
+zenfone
+##ike
+gd
+forum
+jessica
+strong
+alexander
+##ould
+software
+allen
+##ious
+program
+else
+lohasthree
+##gar
+please
+rc
+##ggle
+##ric
+bim
+##own
+eclipse
+brian
+##side
+##other
+##tech
+##ator
+engine
+##ged
+plaza
+##fit
+cia
+ngo
+westbrook
+shi
+tbs
+sci
+reuters
+##ily
+contextlink
+##hn
+af
+##cil
+bridge
+very
+##cel
+cambridge
+##ize
+##aid
+##data
+frm
+##head
+award
+butler
+##sun
+meta
+##mar
+america
+ps3
+puma
+pmid
+lc
+kitchen
+##lic
+day1
+future
+##text
+##page
+##rris
+pm1
+##ket
+fans
+christian
+bot
+kids
+trackback
+##hai
+c3
+display
+##hl
+n2
+idea
+##sent
+airmail
+##ug
+##men
+pwm
+##lution
+awards
+schemas
+asics
+wikipedia
+font
+##tional
+##vy
+c2
+##dget
+##ein
+contact
+pepper
+##uel
+##ument
+##hang
+q5
+##sue
+rain
+##ndi
+wei
+swatch
+##cept
+popular
+##ste
+##tag
+p2
+trc
+##west
+##live
+justin
+honda
+ping
+messenger
+##rap
+v9
+unity
+appqq
+leo
+##tone
+##ass
+uniqlo
+her
+jane
+memory
+moneydj
+##tical
+human
+##m2
+coc
+miacare
+##mn
+tmt
+##core
+vim
+kk
+##may
+fan
+target
+use
+too
+fast
+services
+##ope
+omega
+energy
+pinkoi
+##rain
+jackson
+##ement
+p9
+rd
+##tier
+##vic
+zone
+dl
+isofix
+cpa
+m4
+kimi
+davis
+##lay
+lulu
+##uck
+weeks
+qs
+##hop
+ae
+##ear
+eia
+##fly
+korea
+jpeg
+boost
+##ship
+small
+eur
+valley
+##iel
+simple
+##ude
+rn
+k2
+##ena
+non
+patrick
+feed
+process
+well
+qqmei
+##thing
+they
+aws
+lu
+pink
+##ters
+##kin
+board
+##vertisement
+wine
+##ien
+unicode
+##dge
+r1
+##tant
+##twitter
+cool1
+isp
+standard
+matt
+##fu
+##iner
+googlemsn
+pixnetfacebookyahoo
+x7
+##uce
+sao
+##ev
+##file
+xddd
+shirt
+##rio
+##hat
+givenchy
+ya
+bang
+##lio
+monday
+crystal
+##abc
+head
+ubuntuforumwikilinuxpastechat
+##vc
+##rity
+cnc
+ipv6
+null
+##ost
+yang
+imsean
+tiger
+##fet
+dji
+ji
+maria
+##come
+foundation
+##beth
+active
+##aft
+##don
+sr
+emma
+##khz
+living
+sas
+x6
+##face
+pptv
+x4
+##mate
+han
+sophie
+##jing
+fifa
+##mand
+other
+sale
+inwedding
+##gn
+##mmy
+##pmlast
+bad
+nana
+nbc
+##wu
+note7
+single
+##bel
+window
+##dio
+##ht
+union
+age
+##ivity
+domain
+neo
+##isa
+##lter
+f5
+steven
+##cts
+powerpoint
+tft
+self
+g2
+ft
+zol
+##act
+mwc
+nbapop
+eds
+ace
+##room
+previous
+author
+tomtom
+il
+##ets
+hu
+financial
+bp
+chi
+##hg
+fairmont
+cross
+gay
+h2
+function
+also
+##raph
+##ils
+i3
+avenue
+##host
+##bon
+##tsu
+message
+navigation
+fintech
+h6
+##ject
+##vas
+##firm
+credit
+##wf
+xxxx
+form
+##nor
+##space
+huawei
+plan
+json
+sbl
+##dc
+machine
+wish
+##sol
+windows7
+edward
+development
+washington
+##nsis
+lo
+##sio
+##ym
+##bor
+planet
+##wt
+ieee
+gpa
+camp
+ann
+gm
+##tw
+##oka
+connect
+##rss
+##work
+##atus
+wall
+chicken
+soul
+##times
+fa
+##ather
+##cord
+##eep
+hitachi
+gui
+harry
+##pan
+e1
+disney
+##press
+wind
+frigidaire
+##tl
+liu
+hsu
+basic
+von
+ev
+learning
+##ull
+expedia
+archives
+change
+##wei
+santa
+cut
+ins
+turbo
+brand
+cf1
+return
+##rip
+h1
+##nis
+application
+emc
+rx
+##oon
+quick
+wilson
+wing
+chapter
+##bug
+beyond
+##cms
+##dar
+##oh
+zoom
+e2
+trip
+sb
+##nba
+rcep
+aspx
+ci
+gc
+gnu
+##count
+advanced
+dance
+dv
+##url
+##ging
+am09
+shadow
+battle
+##cia
+emily
+##tation
+host
+ff
+techorz
+sars
+##mini
+##mporary
+##ering
+nc
+##next
+cma
+##mbps
+##gas
+##ift
+##dot
+amana
+##ros
+ir
+##eet
+##ible
+##aka
+dcs
+iq
+l1
+##lor
+maggie
+##iu
+##gt
+articles
+create
+##burg
+##iki
+database
+fantasy
+##rex
+##cam
+dlc
+dean
+##you
+hard
+path
+gaming
+victoria
+maps
+cb
+##lee
+##itor
+overchicstoretvhome
+systems
+##xt
+p3
+sarah
+##nan
+x9
+install
+second
+##ann
+##ph
+##rcle
+##nic
+##nar
+ec
+metro
+chocolate
+##rian
+##table
+skin
+##sn
+mountain
+inparadise
+ib
+##jia
+eeworld
+creative
+g5
+g3
+parker
+ecfa
+village
+sylvia
+hbl
+##ques
+##onsored
+##x2
+##v4
+##tein
+ie6
+##stack
+ver
+##ads
+##baby
+sound
+bbe
+##lone
+##uid
+ads
+gundam
+thinkpad
+scrum
+match
+##ave
+mems
+##oy
+##talk
+glass
+lamigo
+span
+##eme
+job
+##a5
+jay
+wade
+kde
+##lace
+ocean
+tvg
+##covery
+##r3
+##ners
+##rea
+junior
+think
+##aine
+cover
+##ision
+##sia
+##bow
+msi
+##love
+soft
+z2
+##pl
+mobil
+mind
+##uy
+nginx
+##oi
+##rr
+##mple
+##sson
+##nts
+comhd
+crv3000
+##uard
+deep
+lost
+field
+gallery
+##bia
+rate
+spf
+redis
+traction
+icloud
+fe
+jose
+##tory
+into
+sohu
+fx
+kicstart2
+##hia
+##sit
+ra
+##walk
+##xure
+##pact
+pacific
+xa
+natural
+carlo
+##walker
+##can
+cto
+gigi
+pen
+##hoo
+ob
+matlab
+##yy
+##iti
+mango
+##bbs
+sense
+c5
+oxford
+walker
+jennifer
+##ola
+course
+##bre
+##pus
+##rder
+lucky
+ivy
+##nia
+sotheby
+side
+##ugh
+joy
+##orage
+##ush
+##bat
+##dt
+r9
+##gio
+country
+wear
+##lax
+##moon
+seven
+study
+lonzo
+evolution
+##kk
+gs
+kd
+arduino
+b12
+##lux
+arpg
+##rdon
+cook
+##x5
+dark
+five
+##als
+##ida
+sign
+something
+##nda
+##posted
+fresh
+tf
+cam
+##mine
+##skip
+##form
+##ssion
+education
+##tee
+dyson
+stage
+##jie
+want
+##night
+epson
+pack
+##ppy
+##█
+wd
+##eh
+##rence
+left
+##lvin
+golden
+mhz
+discovery
+##trix
+##n2
+loft
+##uch
+##dra
+##sse
+speed
+sorry
+welcome
+##urn
+wave
+gaga
+##lmer
+teddy
+rp
+##sha
+rar
+holiday
+##vg
+##nos
+##rail
+gartner
+gi
+##dium
+kit
+b3
+eco
+sean
+##stone
+autocad
+nu
+##np
+f16
+write
+m5
+##ias
+images
+atp
+##dk
+fsm
+ve
+##xxx
+##cake
+unit
+lim
+ru
+##ification
+published
+angela
+analytics
+ak
+##nel
+gmt
+##icon
+again
+##₂
+##bby
+ios11
+waze
+##ust
+framework
+iptv
+delete
+cl
+wwdc
+##fw
+##xon
+brandt
+##ses
+##dragon
+tc
+vetements
+anne
+monte
+modern
+official
+##ere
+##nne
+##oud
+etnews
+##a2
+##graphy
+##rtex
+l2
+##gma
+mount
+ccd
+archive
+morning
+tan
+ddos
+e7
+day4
+gis
+its
+factory
+bruce
+pg
+##ito
+guest
+cdma
+##lling
+n3
+mega
+eyes
+ro
+women
+dac
+church
+##jun
+singapore
+##facebook
+starbucks
+##tos
+##stin
+##shine
+zen
+##mu
+tina
+request
+##gence
+qt
+q7
+##zzi
+diary
+##tore
+##ead
+cst
+##osa
+canada
+agent
+va
+##jiang
+##lam
+sg
+##nix
+##sday
+g6
+##master
+bing
+##zl
+charlie
+nb40
+thai
+ln284ct
+##itz
+bonnie
+##food
+##lent
+originals
+##stro
+##lts
+##bscribe
+children
+ntd
+yesstyle
+hmv
+##tment
+d5
+arts
+sms
+##pn
+topios9
+lifestyle
+virtual
+##ague
+xz
+##deo
+muji
+unt
+##nnis
+faq1
+##ette
+fly
+curry
+##pop
+release
+##cast
+##ews
+##stle
+ios7
+##ima
+dog
+lenovo
+##r4
+roger
+cbs
+vornado
+##desk
+##ald
+##van
+oil
+some
+break
+common
+##jy
+##lines
+g7
+twice
+ella
+nano
+belle
+##mes
+##self
+##note
+jb
+benz
+##ova
+save
+##wing
+kai
+##hua
+##rect
+rainer
+##unge
+adsl
+guestname
+##uma
+##kins
+##zu
+tokichoi
+##price
+county
+##med
+##mus
+rmk
+address
+vm
+openload
+##group
+##hin
+##iginal
+amg
+urban
+##oz
+jobs
+emi
+##public
+beautiful
+##sch
+album
+##dden
+##bell
+jerry
+works
+hostel
+miller
+##drive
+##rmin
+boot
+##fx
+##nome
+##ctionary
+##oman
+##lish
+##cr
+##hm
+##how
+francis
+xi
+c919
+b5
+evernote
+##uc
+vga
+coupe
+##urg
+##cca
+##uality
+multi
+##ett
+em
+hey
+##ani
+##tax
+##rma
+inside
+than
+leonnhurt
+##jin
+ict
+bird
+notes
+##dical
+##lli
+result
+iu
+ee
+smap
+gopro
+##last
+yin
+pure
+##dan
+##rame
+mama
+##oot
+bean
+marketing
+##hur
+bella
+sync
+xuite
+##ground
+discuz
+##getrelax
+##ince
+##bay
+cj
+gmat
+apt
+##pass
+jing
+##rix
+c4
+rich
+niusnews
+##ello
+bag
+##eting
+##mobile
+culture
+area
+##ience
+details
+gp
+universal
+silver
+dit
+private
+ddd
+u11
+kanshu
+##ified
+fung
+##nny
+dx
+tai
+##fr
+##lean
+##pin
+##rin
+ly
+rick
+##bility
+usb3
+banner
+##baru
+##gion
+metal
+dt
+vdf
+karl
+qualcomm
+bear
+oldid
+ian
+jo
+##tors
+population
+##ernel
+mmorpg
+##mv
+##bike
+ww
+friend
+##ager
+exhibition
+##del
+##pods
+fpx
+structure
+##free
+##tings
+kl
+##rley
+##copyright
+##mma
+california
+orange
+yoga
+canmake
+honey
+##anda
+nikkie
+dhl
+publishing
+##mall
+##gnet
+##┅
+e88
+##dog
+fishbase
+##!
+##"
+###
+##$
+##%
+##&
+##'
+##(
+##)
+##*
+##+
+##,
+##-
+##.
+##/
+##:
+##;
+##<
+##=
+##>
+##?
+##@
+##[
+##\
+##]
+##^
+##_
+##{
+##|
+##}
+##~
+##£
+##¤
+##¥
+##§
+##«
+##±
+##³
+##µ
+##·
+##¹
+##º
+##»
+##¼
+##ß
+##æ
+##÷
+##ø
+##đ
+##ŋ
+##ɔ
+##ə
+##ɡ
+##ʰ
+##ˇ
+##ˈ
+##ˊ
+##ˋ
+##ˍ
+##ː
+##˙
+##˚
+##ˢ
+##α
+##β
+##γ
+##δ
+##ε
+##η
+##θ
+##ι
+##κ
+##λ
+##μ
+##ν
+##ο
+##��
+##ρ
+##ς
+##σ
+##τ
+##υ
+##φ
+##χ
+##ψ
+##б
+##в
+##г
+##д
+##е
+##ж
+##з
+##к
+##л
+##м
+##н
+##о
+##п
+##р
+##с
+##т
+##у
+##ф
+##х
+##ц
+##ч
+##ш
+##ы
+##ь
+##і
+##ก
+##ง
+##น
+##ม
+##ย
+##ร
+##อ
+##า
+##เ
+##๑
+##་
+##ღ
+##ᵃ
+##ᵉ
+##ᵍ
+##ᵏ
+##ᵐ
+##ᵒ
+##ᵘ
+##‖
+##„
+##†
+##•
+##‥
+##‧
+##
+##‰
+##′
+##″
+##‹
+##›
+##※
+##‿
+##⁄
+##ⁱ
+##⁺
+##ⁿ
+##₁
+##₃
+##₄
+##€
+##№
+##ⅰ
+##ⅱ
+##ⅲ
+##ⅳ
+##ⅴ
+##⇒
+##∀
+##−
+##∕
+##∙
+##√
+##∞
+##∟
+##∠
+##∣
+##∩
+##∮
+##∶
+##∼
+##∽
+##≈
+##≒
+##≡
+##≤
+##≥
+##≦
+##≧
+##≪
+##≫
+##⊙
+##⋅
+##⋈
+##⋯
+##⌒
+##①
+##②
+##③
+##④
+##⑤
+##⑥
+##⑦
+##⑧
+##⑨
+##⑩
+##⑴
+##⑵
+##⑶
+##⑷
+##⑸
+##⒈
+##⒉
+##⒊
+##⒋
+##ⓒ
+##ⓔ
+##ⓘ
+##━
+##┃
+##┆
+##┊
+##┌
+##└
+##├
+##┣
+##═
+##║
+##╚
+##╞
+##╠
+##╭
+##╮
+##╯
+##╰
+##╱
+##╳
+##▂
+##▃
+##▅
+##▇
+##▉
+##▋
+##▌
+##▍
+##▎
+##□
+##▬
+##△
+##►
+##▽
+##◇
+##◕
+##◠
+##◢
+##◤
+##☞
+##☼
+##♡
+##♫
+##♬
+##✕
+##✦
+##✪
+##✰
+##✿
+##❀
+##➜
+##➤
+##⦿
+##、
+##。
+##〃
+##々
+##〇
+##〈
+##〉
+##《
+##》
+##「
+##」
+##『
+##』
+##【
+##】
+##〓
+##〔
+##〕
+##〖
+##〗
+##〜
+##〝
+##〞
+##ㄧ
+##ㆍ
+##㈦
+##㊣
+##㗎
+##︰
+##︱
+##︶
+##︿
+##﹁
+##﹂
+##﹍
+##﹏
+##﹐
+##﹑
+##﹒
+##﹔
+##﹕
+##﹖
+##﹗
+##﹙
+##﹚
+##﹝
+##﹞
+##﹡
+##﹣
+##！
+##（
+##）
+##，
+##：
+##；
+##？
+##｡
+##｢
+##｣
+##､
+##･
+##ｯ
+##ｰ
+##ﾞ
+##ﾟ
+##￣
+##￥

utils/zh_util.py

@@ -0,0 +1,91 @@
+"""
+TODO: 繁体、简体
+"""
+import os
+import json
+from collections import Counter
+from utils.text_util import is_chinese, has_chinese
+from zhon.hanzi import punctuation as zh_punc
+
+CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
+
+zh_tokens = [line.strip() for line in open(os.path.join(CURRENT_DIR, "vocab.jd.txt.v2"), "r", encoding="utf-8") if is_chinese(line.strip())]
+
+def zh_iterator():
+    for idx in range(ord(u'\u4e00'), ord(u'\u9fa5')):
+        yield (chr(idx))
+
+
+def get_coding_length(tokenizer, vocab, filter=None):
+    all_length = []
+    for word in vocab:
+        if len(word) > 1:
+            continue
+        if filter is not None and filter(word):
+            continue
+        tokens = tokenizer.encode(word)
+        all_length.append(len(tokens))
+        # if len(tokens.ids) > 1:
+        # if len(tokens) > 3:
+        #     print(word, tokens)
+
+    dist_length = Counter(all_length)
+    mean_length = round(sum(all_length) / len(all_length), 2)
+    return dist_length, mean_length
+
+
+def has_zh_char(text):
+    return any(ch in zh_punc for ch in text)
+
+
+def iter_vocab(tokenizer, name=""):
+    f_out = open(name + "_vocab.zh.jsonl", "w", encoding="utf-8")
+    zh_token_count = {"total": 0, "中文单字": 0, "中文多字": 0}
+    all_single_zh_tokens = set()
+    zh_symbol_count = 0
+    for idx in range(tokenizer.vocab_size):
+        decode_str = tokenizer.decode([idx])
+        if has_chinese(decode_str):
+            # bert词典有 ##开头的
+            # byteBPE词典有带空格的
+            decode_str = decode_str.strip().replace("#", "")
+            zh_token_count["total"] += 1
+            if len(decode_str) > 1:
+                zh_token_count["中文多字"] += 1
+                f_out.write(json.dumps({"id": idx, "token": decode_str, "type": "中文多字"},
+                                       ensure_ascii=False) + "\n")
+            else:
+                all_single_zh_tokens.add(decode_str)
+                zh_token_count["中文单字"] += 1
+                f_out.write(json.dumps({"id": idx, "token": decode_str, "type": "中文单字"},
+                                       ensure_ascii=False) + "\n")
+
+        elif has_zh_char(decode_str):
+            zh_symbol_count += 1
+            f_out.write(json.dumps({"id": idx, "token": decode_str, "type": "中文标点"},
+                                   ensure_ascii=False) + "\n")
+
+    #
+
+    dist_length, mean_length = get_coding_length(tokenizer, zh_tokens, filter=lambda k: not is_chinese(k))
+
+    # TODO: 繁体字，简体字
+    zh_token_count["中文单字-去重后"] = len(all_single_zh_tokens)
+
+    return {
+        "name": name,
+        "impl": str(tokenizer.__class__),
+        "vocab_size": tokenizer.vocab_size,
+        "中文汉字数": str(zh_token_count),
+        "中文标点数": zh_symbol_count,
+        "中文汉字编码长度均值": mean_length,
+        "中文汉字编码长度分布": json.dumps(dist_length),
+
+    }
+
+
+if __name__ == "__main__":
+    # test_coding_length(jd_vocab_tokens, filter=lambda k: not is_chinese(k))
+    # test_coding_length(zh_punc)
+    # test_coding_length(zh_iterator())
+    iter_vocab()

vocab/README.md

@@ -0,0 +1,87 @@
+
+## 词典构建
+
+bert词典
+gpt词典
+gpt-neox词典
+
+## encode
+
+
+## decode
+
+bert词典有个特殊字符 #
+gpt词典有个特殊字符 G
+
+gpt-neox词典呢？
+
+
+## 关于分词粒度
+
+
+## ss
+
+
+
+bert-chinese  vocab_size: 21128
+bert-en
+clue
+glm
+chatglm
+bloom
+
+
+## bert
+
+```
+[PAD]
+...
+[unused99]
+[UNK]
+[CLS]
+[SEP]
+[MASK]
+<S>
+<T>
+!
+...
+
+big
+##ut
+ftp
+carol
+##vi
+```
+
+
+## 
+
+https://github.com/pytorch/fairseq/blob/master/tests/test_noising.py#L37
+
+```
+"he@@", "llo", "n@@", "ew", "y@@", "or@@", "k"
+```
+
+跟BERT类似，只不过BERT是词后缀，这里是词前缀。
+
+
+## GPT2
+
+词典见：https://huggingface.co/gpt2/raw/main/vocab.json
+
+
+```
+['What', "'s", 'Ġup', 'Ġwith', 'Ġthe', 'Ġtoken', 'izer', '?']
+```
+跟BERT不同，BERT用特殊符号表示 “连接”，GPT2用特殊符号表示 “空格”。
+
+详见 gpt2/README.md
+
+- 功能符号： `<|endoftext|>` 表示换行。tab？ 空格？
+- 很多数字独立编码，几乎上千个。
+
+## 空格、tab、换行 
+
+
+
+## 

vocab/__init__.py

@@ -0,0 +1,103 @@
+import transformers
+import importlib
+from enum import Enum, auto
+
+Animal = Enum('Animal', 'ANT BEE CAT DOG')
+
+uniq_tokenizers = [
+    ""
+]
+
+all_tokenizers = [
+    "gpt_35_turbo",
+    "gpt2",
+    "gpt2_chinese",
+    "bert_chinese",
+    "moss",
+    #
+    # ######
+    # "chatyuan_large_v2",
+    # "prompt_clue",
+    #
+    # #### bloom 系列
+    # "bloom",
+    # "bloomz_6b4_zh",
+    # "belle_7b_2m",   # 模型和词典都基于bloom
+    #
+    "gpt_nexo_20b",
+    # "gpt_neox_chinese_v1",
+    #
+    # ##### glm系列
+    # "glm_chinese",
+    "chatglm",
+    #
+    # #### llama alpaca系列
+    "llama",  #  '中文单字': 700, '中文多字': 0
+    "chinese_llama_lora_7b",  #
+    # "chinese_alpaca_lora_7b",  # 中文Alpaca模型在上述中文LLaMA模型的基础上进一步使用了指令数据进行精调。
+    # "belle_llama_ext_7b",
+    # "alpaca_7b",
+    "baichuan_7b",
+    "qwen"
+]
+
+
+
+class TokenizerType(Enum):
+    """
+    - https://huggingface.co/docs/transformers/tokenizer_summary
+    - https://github.com/EleutherAI/gpt-neox/blob/main/megatron/tokenizer/tokenizer.py
+    - https://github.com/google/sentencepiece/blob/3863f7648e5d8edb571ac592f3ac4f5f0695275a/src/sentencepiece_model.proto#L48
+      - UNIGRAM = 1;  // Unigram language model with dynamic algorithm
+      - BPE = 2;      // Byte Pair Encoding
+      - WORD = 3;     // Delimitered by whitespace.
+      - CHAR = 4;     // tokenizes into character sequence
+    """
+    BPE = auto()
+    ByteBPE = auto()  # BBPE  Byte-Level BPE
+    GPT2BPETokenizer = auto()  #
+    BERTTokenizer = auto()
+
+
+# class TokenizerType(Enum):
+#
+#     # BERTTokenizer
+#     # 依赖一个txt文件
+#
+#
+#     # https://github.com/EleutherAI/gpt-neox/blob/v2.0/megatron/tokenizer/tokenizer.py#L231
+#     # 依赖一个json文件，Tokenizer.from_file(vocab_file)
+#     # 案例：gpt-neox-20B
+#     HFTokenizer = auto()
+#
+#     # 依赖: model_file, sentencepiece.SentencePieceProcessor(model_file)
+#     # 案例：
+#     SentencePieceTokenizer = auto()
+#
+#
+#     # 依赖: 3个json文件：vocab.json, merges.txt, special_tokens.txt
+#     # 源码:
+#     #   - https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/tokenizer/gpt2_tokenization.py#L92
+#     # Byte-level BPE
+#     GPT2BPETokenizer = auto()
+
+
+class TokenizerImpl(Enum):
+    """
+    """
+    SentencePiece = auto()  #
+
+    # https://github.com/huggingface/transformers/blob/v4.30.2/src/transformers/models/gpt2/tokenization_gpt2.py#L104
+    # 构造词典：
+    #
+    GPT2Tokenizer = auto()
+    BertTokenizer = auto()  #
+
+
+
+def load_tokener(model_name):
+    tokenizer = importlib.import_module("." + model_name, 'vocab').tokenizer
+    return tokenizer
+
+if __name__ == "__main__":
+    pass

vocab/albert/__init__.py

@@ -0,0 +1,6 @@
+"""
+
+SentencePiece(unigram)
+
+https://huggingface.co/docs/transformers/tokenizer_summary#sentencepiece
+"""

vocab/baichuan_7b/Baichuan-7B/config.json

@@ -0,0 +1,26 @@
+{
+  "architectures": [
+    "BaiChuanForCausalLM"
+  ],
+  "auto_map": {
+    "AutoConfig": "configuration_baichuan.BaiChuanConfig",
+    "AutoModelForCausalLM": "modeling_baichuan.BaiChuanForCausalLM"
+  },
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 4096,
+  "model_type": "baichuan",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": "4.29.1",
+  "use_cache": true,
+  "vocab_size": 64000
+}

vocab/baichuan_7b/Baichuan-7B/configuration_baichuan.py

@@ -0,0 +1,66 @@
+# coding=utf-8
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+
+class BaiChuanConfig(PretrainedConfig):
+    model_type = "baichuan"
+    keys_to_ignore_at_inference = ["past_key_values"]
+
+    def __init__(
+        self,
+        vocab_size=64000,
+        hidden_size=4096,
+        intermediate_size=11008,
+        num_hidden_layers=32,
+        num_attention_heads=32,
+        hidden_act="silu",
+        max_position_embeddings=4096,
+        initializer_range=0.02,
+        rms_norm_eps=1e-6,
+        use_cache=True,
+        pad_token_id=0,
+        bos_token_id=1,
+        eos_token_id=2,
+        tie_word_embeddings=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.max_position_embeddings = max_position_embeddings
+        self.hidden_size = hidden_size
+        self.intermediate_size = intermediate_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.hidden_act = hidden_act
+        self.initializer_range = initializer_range
+        self.rms_norm_eps = rms_norm_eps
+        self.use_cache = use_cache
+        super().__init__(
+            pad_token_id=pad_token_id,
+            bos_token_id=bos_token_id,
+            eos_token_id=eos_token_id,
+            tie_word_embeddings=tie_word_embeddings,
+            **kwargs,
+        )

vocab/baichuan_7b/Baichuan-7B/special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

vocab/baichuan_7b/Baichuan-7B/tokenization_baichuan.py

@@ -0,0 +1,250 @@
+# coding=utf-8
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import os
+from shutil import copyfile
+from typing import Any, Dict, List, Optional, Tuple
+
+import sentencepiece as spm
+
+from transformers.tokenization_utils import AddedToken, PreTrainedTokenizer
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+VOCAB_FILES_NAMES = {"vocab_file": "tokenizer.model"}
+
+PRETRAINED_VOCAB_FILES_MAP = {
+    "vocab_file": {},
+    "tokenizer_file": {},
+}
+PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES = {}
+
+
+class BaiChuanTokenizer(PreTrainedTokenizer):
+    """
+    Construct a BaiChuan tokenizer. Based on byte-level Byte-Pair-Encoding.
+
+    Args:
+        vocab_file (`str`):
+            Path to the vocabulary file.
+    """
+
+    vocab_files_names = VOCAB_FILES_NAMES
+    pretrained_vocab_files_map = PRETRAINED_VOCAB_FILES_MAP
+    max_model_input_sizes = PRETRAINED_POSITIONAL_EMBEDDINGS_SIZES
+    model_input_names = ["input_ids", "attention_mask"]
+
+    def __init__(
+        self,
+        vocab_file,
+        unk_token="<unk>",
+        bos_token="<s>",
+        eos_token="</s>",
+        pad_token=None,
+        sp_model_kwargs: Optional[Dict[str, Any]] = None,
+        add_bos_token=True,
+        add_eos_token=False,
+        clean_up_tokenization_spaces=False,
+        **kwargs,
+    ):
+        self.sp_model_kwargs = {} if sp_model_kwargs is None else sp_model_kwargs
+        bos_token = AddedToken(bos_token, lstrip=False, rstrip=False) if isinstance(bos_token, str) else bos_token
+        eos_token = AddedToken(eos_token, lstrip=False, rstrip=False) if isinstance(eos_token, str) else eos_token
+        unk_token = AddedToken(unk_token, lstrip=False, rstrip=False) if isinstance(unk_token, str) else unk_token
+        pad_token = AddedToken(pad_token, lstrip=False, rstrip=False) if isinstance(pad_token, str) else pad_token
+        super().__init__(
+            bos_token=bos_token,
+            eos_token=eos_token,
+            unk_token=unk_token,
+            pad_token=pad_token,
+            add_bos_token=add_bos_token,
+            add_eos_token=add_eos_token,
+            sp_model_kwargs=self.sp_model_kwargs,
+            clean_up_tokenization_spaces=clean_up_tokenization_spaces,
+            **kwargs,
+        )
+        self.vocab_file = vocab_file
+        self.add_bos_token = add_bos_token
+        self.add_eos_token = add_eos_token
+        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
+        self.sp_model.Load(vocab_file)
+
+    def __getstate__(self):
+        state = self.__dict__.copy()
+        state["sp_model"] = None
+        return state
+
+    def __setstate__(self, d):
+        self.__dict__ = d
+        self.sp_model = spm.SentencePieceProcessor(**self.sp_model_kwargs)
+        self.sp_model.Load(self.vocab_file)
+
+    @property
+    def vocab_size(self):
+        """Returns vocab size"""
+        return self.sp_model.get_piece_size()
+
+    def get_vocab(self):
+        """Returns vocab as a dict"""
+        vocab = {self.convert_ids_to_tokens(i): i for i in range(self.vocab_size)}
+        vocab.update(self.added_tokens_encoder)
+        return vocab
+
+    def _tokenize(self, text):
+        """Returns a tokenized string."""
+        return self.sp_model.encode(text, out_type=str)
+
+    def _convert_token_to_id(self, token):
+        """Converts a token (str) in an id using the vocab."""
+        return self.sp_model.piece_to_id(token)
+
+    def _convert_id_to_token(self, index):
+        """Converts an index (integer) in a token (str) using the vocab."""
+        token = self.sp_model.IdToPiece(index)
+        return token
+
+    def convert_tokens_to_string(self, tokens):
+        """Converts a sequence of tokens (string) in a single string."""
+        current_sub_tokens = []
+        out_string = ""
+        prev_is_special = False
+        for i, token in enumerate(tokens):
+            # make sure that special tokens are not decoded using sentencepiece model
+            if token in self.all_special_tokens:
+                if not prev_is_special and i != 0:
+                    out_string += " "
+                out_string += self.sp_model.decode(current_sub_tokens) + token
+                prev_is_special = True
+                current_sub_tokens = []
+            else:
+                current_sub_tokens.append(token)
+                prev_is_special = False
+        out_string += self.sp_model.decode(current_sub_tokens)
+        return out_string
+
+    def save_vocabulary(self, save_directory, filename_prefix: Optional[str] = None) -> Tuple[str]:
+        """
+        Save the vocabulary and special tokens file to a directory.
+
+        Args:
+            save_directory (`str`):
+                The directory in which to save the vocabulary.
+
+        Returns:
+            `Tuple(str)`: Paths to the files saved.
+        """
+        if not os.path.isdir(save_directory):
+            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
+            return
+        out_vocab_file = os.path.join(
+            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
+        )
+
+        if os.path.abspath(self.vocab_file) != os.path.abspath(out_vocab_file) and os.path.isfile(self.vocab_file):
+            copyfile(self.vocab_file, out_vocab_file)
+        elif not os.path.isfile(self.vocab_file):
+            with open(out_vocab_file, "wb") as fi:
+                content_spiece_model = self.sp_model.serialized_model_proto()
+                fi.write(content_spiece_model)
+
+        return (out_vocab_file,)
+
+    def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
+        bos_token_id = [self.bos_token_id] if self.add_bos_token else []
+        eos_token_id = [self.eos_token_id] if self.add_eos_token else []
+
+        output = bos_token_id + token_ids_0 + eos_token_id
+
+        if token_ids_1 is not None:
+            output = output + bos_token_id + token_ids_1 + eos_token_id
+
+        return output
+
+    def get_special_tokens_mask(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None, already_has_special_tokens: bool = False
+    ) -> List[int]:
+        """
+        Retrieve sequence ids from a token list that has no special tokens added. This method is called when adding
+        special tokens using the tokenizer `prepare_for_model` method.
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of IDs.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
+                Whether or not the token list is already formatted with special tokens for the model.
+
+        Returns:
+            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
+        """
+        if already_has_special_tokens:
+            return super().get_special_tokens_mask(
+                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
+            )
+
+        bos_token_id = [1] if self.add_bos_token else []
+        eos_token_id = [1] if self.add_eos_token else []
+
+        if token_ids_1 is None:
+            return bos_token_id + ([0] * len(token_ids_0)) + eos_token_id
+        return (
+            bos_token_id
+            + ([0] * len(token_ids_0))
+            + eos_token_id
+            + bos_token_id
+            + ([0] * len(token_ids_1))
+            + eos_token_id
+        )
+
+    def create_token_type_ids_from_sequences(
+        self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None
+    ) -> List[int]:
+        """
+        Creates a mask from the two sequences passed to be used in a sequence-pair classification task. An ALBERT
+        sequence pair mask has the following format:
+
+        ```
+        0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1
+        | first sequence    | second sequence |
+        ```
+
+        if token_ids_1 is None, only returns the first portion of the mask (0s).
+
+        Args:
+            token_ids_0 (`List[int]`):
+                List of ids.
+            token_ids_1 (`List[int]`, *optional*):
+                Optional second list of IDs for sequence pairs.
+
+        Returns:
+            `List[int]`: List of [token type IDs](../glossary#token-type-ids) according to the given sequence(s).
+        """
+        bos_token_id = [self.bos_token_id] if self.add_bos_token else []
+        eos_token_id = [self.eos_token_id] if self.add_eos_token else []
+
+        output = [0] * len(bos_token_id + token_ids_0 + eos_token_id)
+
+        if token_ids_1 is not None:
+            output += [1] * len(bos_token_id + token_ids_1 + eos_token_id)
+
+        return output

vocab/baichuan_7b/Baichuan-7B/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4be54af290d93c113bcbf421115ae9eed9d6340408f564898f1e966dc738ef01
+size 1136699

vocab/baichuan_7b/Baichuan-7B/tokenizer_config.json

@@ -0,0 +1,35 @@
+{
+  "auto_map": {
+    "AutoTokenizer": ["tokenization_baichuan.BaiChuanTokenizer", null]
+  },
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 1000000000000000019884624838656,
+  "sp_model_kwargs": {},
+  "tokenizer_class": "BaiChuanTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

vocab/baichuan_7b/__init__.py

@@ -0,0 +1,8 @@
+from transformers import AutoTokenizer
+from vocab import TokenizerType
+
+tokenizer = AutoTokenizer.from_pretrained("baichuan-inc/Baichuan-7B", trust_remote_code=True)
+
+
+# byte-bpe  sentencepiece
+tokenizer.type = TokenizerType.ByteBPE

vocab/belle_7b_2m/README.md

@@ -0,0 +1,3 @@
+
+采用的bloom词典
+

vocab/belle_7b_2m/__init__.py

@@ -0,0 +1,8 @@
+
+import os
+from transformers import AutoTokenizer
+
+CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
+TOKENIZER_DIR = os.path.join(CURRENT_DIR, "belle-7b-2m")
+
+tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_DIR)

vocab/belle_7b_2m/belle-7b-2m/README.md

@@ -0,0 +1,3 @@
+
+
+来自 https://huggingface.co/BelleGroup/BELLE-7B-2M

vocab/belle_7b_2m/belle-7b-2m/config.json

@@ -0,0 +1,31 @@
+{
+  "apply_residual_connection_post_layernorm": false,
+  "architectures": [
+    "BloomModel"
+  ],
+  "attention_dropout": 0.0,
+  "attention_softmax_in_fp32": true,
+  "bias_dropout_fusion": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_dropout": 0.0,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "masked_softmax_fusion": true,
+  "model_type": "bloom",
+  "n_embed": 4096,
+  "n_inner": null,
+  "n_layer": 30,
+  "num_attention_heads": 32,
+  "offset_alibi": 100,
+  "pad_token_id": 3,
+  "pretraining_tp": 4,
+  "seq_length": 2048,
+  "skip_bias_add": true,
+  "skip_bias_add_qkv": false,
+  "slow_but_exact": false,
+  "transformers_version": "4.21.0.dev0",
+  "unk_token_id": 0,
+  "use_cache": true,
+  "vocab_size": 250880
+}

vocab/belle_7b_2m/belle-7b-2m/special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

vocab/belle_7b_2m/belle-7b-2m/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fa39cd4b1500feb205bcce3b9703a4373414cafe4970e0657b413f7ddd2a9d3
+size 14500438

vocab/belle_7b_2m/belle-7b-2m/tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "eos_token": "</s>", "bos_token": "<s>", "pad_token": "<pad>", "name_or_path": "bigscience/tokenizer", "special_tokens_map_file": null, "tokenizer_class": "BloomTokenizerFast"}

vocab/belle_llama_ext_7b/BELLE-LLaMA-EXT-7B/config.json

@@ -0,0 +1,22 @@
+{
+  "_name_or_path": "/nfs/a100-80G-17/jiyunjie/finetuned_ckpt/hf_llama_7b_expand_belle_tokenizer50k/wudao_200G_pretrain_filter_len2000_instr_l512_epoch=0-step=6400",
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.28.0.dev0",
+  "use_cache": true,
+  "vocab_size": 79458
+}

vocab/belle_llama_ext_7b/BELLE-LLaMA-EXT-7B/generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "pad_token_id": 0,
+  "transformers_version": "4.28.0.dev0"
+}

vocab/belle_llama_ext_7b/BELLE-LLaMA-EXT-7B/special_tokens_map.json

@@ -0,0 +1,5 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "unk_token": "<unk>"
+}

vocab/belle_llama_ext_7b/BELLE-LLaMA-EXT-7B/tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b716a618c9e7c45648f91d997431eba3b0ff111b17ce7b777280ed771a49f95
+size 1268137

vocab/belle_llama_ext_7b/BELLE-LLaMA-EXT-7B/tokenizer_config.json

@@ -0,0 +1,8 @@
+{
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "special_tokens_map_file": "/nfs/v100-022/pretrained_ckpt/hf_llama_7b_expand_belle_tokenizer50k_only/special_tokens_map.json",
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": "<unk>"
+}

vocab/belle_llama_ext_7b/__init__.py

@@ -0,0 +1,12 @@
+
+import os
+from transformers import LlamaTokenizer
+
+CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
+TOKENIZER_DIR = os.path.join(CURRENT_DIR, "BELLE-LLaMA-EXT-7B")
+
+tokenizer = LlamaTokenizer.from_pretrained(TOKENIZER_DIR)
+
+
+
+# print(tokenizer.tokenize("大道发生"))

vocab/bert_chinese/README.md

@@ -0,0 +1,2 @@
+
+vocab_size: 21128

vocab/bert_chinese/__init__.py

@@ -0,0 +1,11 @@
+
+import os
+from transformers import AutoTokenizer
+
+tokenizer = AutoTokenizer.from_pretrained("bert-base-chinese")
+
+
+
+
+# vocab_size = len(tokenizer.get_vocab())
+# vocab_size = tokenizer.vocab_size

vocab/bert_chinese/test.py

@@ -0,0 +1,6 @@
+
+from transformers import AutoTokenizer, AutoModelForMaskedLM
+tokenizer = AutoTokenizer.from_pretrained("bert-base-chinese")
+
+tokens = tokenizer.tokenize("明天搜索")
+print(tokens)

vocab/bert_chinese/test_zh_coding_len.py

@@ -0,0 +1,73 @@
+"""
+1. jd_vocab_tokens的中文：
+
+
+
+2. 中文标点
+
+
+3. 全中文（单字） unicode
+
+
+4. 全中文（）
+中文汉字数：54230, 中文标点数: 549
+"""
+
+from collections import Counter
+from transformers import AutoTokenizer
+from data_sample.oov_base import jd_vocab_tokens
+from utils.text_util import is_chinese, has_chinese
+from zhon.hanzi import punctuation as zh_punc
+
+
+vocab = open("vocab.txt", "r", encoding="utf-8").readlines()
+
+def zh_iterator():
+    for idx in range(ord(u'\u4e00'), ord(u'\u9fa5')):
+        yield (chr(idx))
+
+
+def test_coding_length(vocab, filter=None):
+    all_length = []
+    for word in vocab:
+        if len(word) > 1:
+            continue
+        if filter is not None and filter(word):
+            continue
+        tokens = tokenizer.encode(word)
+        all_length.append(len(tokens))
+        # if len(tokens.ids) > 1:
+        if len(tokens.ids) == 1:
+            print(word, tokens.ids)
+
+    print("编码长度统计：", Counter(all_length))
+    print("平均编码长度：", sum(all_length)/len(all_length))
+
+
+def has_zh_char(text):
+    return any(ch in zh_punc for ch in text)
+
+
+def iter_vocab():
+
+    f_out = open("vocab.zh.txt", "w", encoding="utf-8")
+    zh_token_count = 0
+    zh_symbol_count = 0
+    for idx, word in enumerate(vocab):
+
+        if has_chinese(decode_str):
+            zh_token_count += 1
+            f_out.write("%d\t%s\t中文汉字\n" % (idx, decode_str))
+        elif has_zh_char(decode_str):
+            zh_symbol_count += 1
+            f_out.write("%d\t%s\t中文标点\n" % (idx, decode_str))
+
+    print("中文汉字数：%d, 中文标点数: %d" % (zh_token_count, zh_symbol_count))
+
+
+if __name__ == "__main__":
+
+    # test_coding_length(jd_vocab_tokens, filter=lambda k: not is_chinese(k))
+    # test_coding_length(zh_punc)
+    # test_coding_length(zh_iterator())
+    iter_vocab()

vocab/bert_chinese/tokenizer/config.json

@@ -0,0 +1,25 @@
+{
+  "architectures": [
+    "BertForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "directionality": "bidi",
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "bert",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_fc_size": 768,
+  "pooler_num_attention_heads": 12,
+  "pooler_num_fc_layers": 3,
+  "pooler_size_per_head": 128,
+  "pooler_type": "first_token_transform",
+  "type_vocab_size": 2,
+  "vocab_size": 21128
+}

vocab/bert_chinese/tokenizer/tokenizer_config.json

@@ -0,0 +1,3 @@
+{
+  "do_lower_case": false
+}

vocab/bert_clue/README.md

@@ -0,0 +1,7 @@
+
+删除了无用字符，词典更小。增加了少部分
+
+
+## 词典大小
+
+- 

vocab/bert_clue/clue_vocab.py

@@ -0,0 +1,188 @@
+"""
+
+"""
+
+import codecs
+import sys
+import re
+from langconv import *
+import emoji
+
+emoji_regex = emoji.get_emoji_regexp()
+
+human_list = ['▲top', '▲topoct', '▲topmay', '▲topapr', '▲topmar', '▲topjun', '▲topdec', '▲topnov', '▲topaug', '▲topjul',
+              '▲topjan', '▲topsep', '▲topfeb', '￥799', '￥2899', '～～', '～～～', '##～6', '##～10', '～10', '##～5', '～5',
+              '##～20', '##～8', '##～17', '##～1', '～4', '##～3', '##～7', '～1', 'ｗedding', '×email', 'ｃｐ', '××', 'ｏｋ', 'ａ',
+              'ｂ', 'ｃ', 'ｄ', 'ｅ', 'ｆ', 'ｇ', 'ｈ', 'ｉ', 'ｊ', 'ｋ', 'ｌ', 'ｍ', 'ｎ', 'ｏ', 'ｐ', 'ｑ', 'ｒ', 'ｓ', 'ｔ', 'ｕ', 'ｖ',
+              'ｗ', 'ｘ', 'ｙ', 'ｚ', '##★', '##℃', '##～', '##°', '##☆', '↓↓↓', '##●', '##㎡', '##♪', '##×', '▌♥', '##｜',
+              '##ｄ', '##▲', '##ｏ', '★★', '##→', '#ａ', '⋯⋯', '##▼', '##○', '★★★★★', '##∥', '##◆', '##ω', '★★★', '##ｃ',
+              '##ｓ', '##ｅ', '##ｐ', '##■', '##↑', '##ｋ', '##и', '◆◆', '##ｇ', '##＋', '##а', '±0', '##◎', '##─', '##ｒ',
+              '##＞', '##²', '##ｔ', '★★★★', '##│', '##ｎ', '##ｌ', '##＝', '##ｙ', '☆☆☆', '##ｉ', '##↓', 'ˋ▽ˊ', '##ｖ', '↓↓',
+              '##f2016', '##ｑ', '##₂', '∟∣', '##я', '##←', '##◆◆', '##ｘ', '##cm～', '##ｆ', '##ｈ', '##ｊ', '##ｕ', '##ｗ',
+              '##ｚ']
+
+zhuyin_char = ['ㄅ', 'ㄆ', 'ㆠ', 'ㄇ', 'ㄈ', 'ㄪ', 'ㄉ', 'ㄊ', 'ㄋ', 'ㆹ', 'ㄌ', 'ㄍ', 'ㄎ', 'ㆣ', 'ㄫ', 'ㄏ', 'ㆸ', 'ㄐ', 'ㄑ', 'ㆢ', 'ㄬ',
+               'ㄒ', 'ㆺ', 'ㄓ', 'ㄔ', 'ㄕ', 'ㄖ', 'ㄗ', 'ㄘ', 'ㆡ', 'ㄙ', 'ㆡ', 'ㆪ', 'ㄨ', 'ㆫ', 'ㆨ', 'ㄩ', 'ㄚ', 'ㆩ', 'ㆦ', 'ㆧ', 'ㄛ',
+               'ㄜ', 'ㄝ', 'ㆤ', 'ㆥ', 'ㄞ', 'ㆮ', 'ㄟ', 'ㄠ', 'ㆯ', 'ㄡ', 'ㆰ', 'ㆱ', 'ㆬ', 'ㄢ', 'ㄣ', 'ㄯ', 'ㄤ', 'ㆲ', 'ㄥ', 'ㆭ', 'ㄦ',
+               'ㄭ']
+
+special_token = ['[PAD]', '[UNK]', '[CLS]', '[SEP]', '[MASK]', '<S>', '<T>']
+
+japan_chars = ['ｲ', 'ｸ', 'ｼ', 'ｽ', 'ﾄ', 'ﾉ', 'ﾌ', 'ﾗ', 'ﾙ', 'ﾝ']
+
+korean_chars = ['ᄀ', 'ᄁ', 'ᄂ', 'ᄃ', 'ᄅ', 'ᄆ', 'ᄇ', 'ᄈ', 'ᄉ', 'ᄋ', 'ᄌ', 'ᄎ', 'ᄏ', 'ᄐ', 'ᄑ', 'ᄒ', 'ᅡ', 'ᅢ', 'ᅣ', 'ᅥ', 'ᅦ',
+                'ᅧ', 'ᅨ', 'ᅩ', 'ᅪ', 'ᅬ', 'ᅭ', 'ᅮ', 'ᅯ', 'ᅲ', 'ᅳ', 'ᅴ', 'ᅵ', 'ᆨ', 'ᆫ', 'ᆯ', 'ᆷ', 'ᆸ', 'ᆺ', 'ᆻ', 'ᆼ', 'ᗜ']
+
+puns = ['“', '‘']
+
+nums = ['10', '11', '12', '13', '14', '15', '16', '17', '18', '19', '20', '21', '22', '23', '24', '25', '26', '27',
+        '28', '29', '30', '31', '32', '33', '34', '35', '36', '37', '38', '39', '40', '41', '42', '43', '44', '45',
+        '46', '47', '48', '49', '50', '51', '52', '53', '54', '55', '56', '57', '58', '59', '60', '61', '62', '63',
+        '64', '65', '66', '67', '68', '69', '70', '71', '72', '73', '74', '75', '76', '77', '78', '79', '80', '81',
+        '82', '83', '84', '85', '86', '87', '88', '89', '90', '91', '92', '93', '94', '95', '96', '97', '98', '99',
+        '100', '2016', '2017', '2018', '2019', '2020', '2021', '2022']
+
+with open('vocab.google.txt', 'r', encoding='utf8') as fin, \
+        open('vocab.clue.txt', 'w', encoding='utf8') as fout:
+    cout_zh = 0
+    cout_en = 0
+    cout_jp = 0
+    cout_em = 0
+    cout_zh_res = 0
+    cout_zh_tra = 0
+    cout_zh_wp = 0
+    cout_en_del = 0
+    cout_en_res = 0
+    cout_num = 0
+    cout_num_del = 0
+    cout_num_res = 0
+    cout_hand_del = 0
+    cout_total = 0
+    cout_zhuyin = 0
+    cout_unused = 0
+    cout_special = 0
+    cout_jp = 0
+    cout_ko = 0
+
+    for line in fin:
+        cout_total += 1
+        token = line.strip()
+
+        if not token:
+            continue
+
+        if token in human_list:
+            cout_hand_del += 1  # 13
+            continue
+
+        # chinese character
+        elif re.match(u'[\u4e00-\u9fa5]+', token.replace('##', '')):
+            cout_zh += 1  # 14642
+
+            token_simp = Converter('zh-hans').convert(token)
+            if token_simp != token:
+                cout_zh_tra += 1
+                continue
+            else:
+                if re.match(u'##', token):
+                    # print(token)
+                    cout_zh_wp += 1
+                    continue
+                else:
+                    cout_zh_res += 1
+                    print(token, file=fout)
+
+        # korean character
+        elif re.match(u'[\uac00-\ud7ff]+', token.replace('##', '')):
+            # print(token)
+            cout_ko += 1
+            continue
+
+        # japanese character
+        elif re.match(u'[\u30a0-\u30ff\u3040-\u309f]+', token.replace('##', '')):
+            # print(token)
+            cout_jp += 1
+            continue
+
+        # english character
+        elif re.match(u'[a-z]+', token.replace('##', '')):
+            # print(token)
+            cout_en += 1
+            if re.match(u'##', token):
+                # print(token)
+                cout_en_res += 1
+                print(token, file=fout)
+            elif len(token) > 1:
+                # print(token)
+                cout_en_del += 1
+                continue
+            else:
+                # print(token)
+                cout_en_res += 1
+                print(token, file=fout)
+
+        # emoji character
+        elif re.match(emoji_regex, token.replace('##', '')):
+            # print(token)
+            cout_em += 1
+            continue
+
+        # multi-number characters
+        elif re.match(u'(##)?\d', token):
+            cout_num += 1
+            if len(token.replace('##', '')) == 1:
+                # print(token)
+                cout_num_res += 1
+                print(token, file=fout)
+            else:
+                cout_num_del += 1
+                # print(token)
+                continue
+        elif token.replace('##', '') in zhuyin_char:
+            # print(token, file=fout)
+            cout_zhuyin += 1
+            continue
+        elif token.startswith('[unused'):
+            print(token, file=fout)
+            cout_unused += 1
+        elif token in special_token:
+            print(token, file=fout)
+            cout_special += 1
+
+        elif token.replace('##', '') in japan_chars:
+            cout_jp += 1
+            continue
+
+        elif token.replace('##', '') in korean_chars:
+            cout_ko += 1
+            continue
+        else:
+            # print(token)
+            print(token, file=fout)
+
+        # add tokens
+        if token == '"':
+            for token in puns:
+                print(token, file=fout)
+        if token == '9':
+            for token in nums:
+                cout_num_res += 1
+                print(token, file=fout)
+
+    print("cout_zh:{}".format(cout_zh))  # 14642
+    print("cout_zh_tra:{}".format(cout_zh_tra))  # 3264
+    print("cout_zh_wp:{}".format(cout_zh_wp))  # 5689
+    print("cout_zh_res:{}".format(cout_zh_res))  # 5689
+    print("cout_en:{}".format(cout_en))  # 3555
+    print("cout_en_del:{}".format(cout_en_del))  # 2235
+    print("cout_en_res:{}".format(cout_en_res))  # 1320
+    print("cout_num:{}".format(cout_num))  # 1179
+    print("cout_num_del:{}".format(cout_num_del))  # 1137
+    print("cout_num_res:{}".format(cout_num_res))  # 140
+    print("cout_hand_del:{}".format(cout_hand_del))  # 132
+    print("cout_zhuyin:{}".format(cout_zhuyin))  # 36
+    print("cout_unused:{}".format(cout_unused))  # 99
+    print("cout_special:{}".format(cout_special))  # 7
+    print("cout_jp:{}".format(cout_jp))  # 573
+    print("cout_ko:{}".format(cout_ko))  # 84
+    print("cout_em:{}".format(cout_em))  # 56