xianyu564 commited on
Commit
6009933
·
1 Parent(s): 28a443d

更新 requirements.txt,添加新依赖项以支持最新功能

Browse files
data/dataset_stats.json ADDED
@@ -0,0 +1,369 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "total_articles": 317,
3
+ "total_samples": 3030,
4
+ "train_samples": 2727,
5
+ "val_samples": 303,
6
+ "val_split": 0.1,
7
+ "years": [
8
+ "2017",
9
+ "2018",
10
+ "2019",
11
+ "2020",
12
+ "2021",
13
+ "2022",
14
+ "2023",
15
+ "2024",
16
+ "2025",
17
+ "unknown"
18
+ ],
19
+ "topics": [
20
+ "2017-07-31_破蛋推_衔瑜今天成年啦!!!",
21
+ "2017-08-01_衔瑜小札_2017.07.31",
22
+ "2017-08-02_衔瑜攻略_2017.08.02",
23
+ "2017-08-03_衔瑜文艺_好久不见",
24
+ "2017-08-04_衔瑜训诂_说尬",
25
+ "2017-08-06_衔瑜脑洞_五条衔瑜 鱼目混珠",
26
+ "2017-08-08_公号写了一周了 可是人设还没有出来",
27
+ "2017-08-11_杂感",
28
+ "2017-08-20_留下尬走的足迹_杭州",
29
+ "2017-08-22_留下尬走的足迹_苏州",
30
+ "2017-08-23_留下尬走的足迹_上海",
31
+ "2017-08-25_留下的尬走足迹_大连",
32
+ "2017-09-01_留下的尬走足迹_香港",
33
+ "2017-09-17_子龙一身都是胆 我一身都是肝",
34
+ "2017-09-21_奶茶要不啦~奶茶",
35
+ "2017-09-24_一周之内 而气候不齐",
36
+ "2017-10-02_化学生的求生攻略",
37
+ "2017-10-05_从狼人杀看《李尔王》",
38
+ "2017-10-06_几件小事",
39
+ "2017-10-21_想讨论点关于教育的事",
40
+ "2017-10-29_在油腻的生活中诗意地栖居",
41
+ "2017-11-08_无以言表的集体感",
42
+ "2017-11-10_今年双十一也没点长进吗",
43
+ "2017-11-11_在不安的世界里 安静地活着",
44
+ "2017-11-16_雅礼,今天我为你写下小札",
45
+ "2017-11-21_猜猜我今天要说什么",
46
+ "2017-11-30_别到了长沙 还不知道怎么嗦粉",
47
+ "2017-12-15_今天读篇鲁迅 说点有的没的",
48
+ "2017-12-23_苦尽肝来",
49
+ "2017-12-25_君子之交肝若醴",
50
+ "2017-12-28_这是你的羁绊",
51
+ "2018-01-01_今天这一篇是凑数的",
52
+ "2018-01-05_日子就很平常",
53
+ "2018-01-06_舍友都出门了的时候",
54
+ "2018-01-13_写在大物草稿纸上的小说",
55
+ "2018-01-19_回归·失踪人口·张衔瑜",
56
+ "2018-01-20_今晚看话剧回来 心里像憋着一壶叫喊",
57
+ "2018-01-22_只是尽可能的简单 却不再简化",
58
+ "2018-01-31_今天我助力了天朝的老龄化进程",
59
+ "2018-01-31_姗姗来迟的『茶颜悦色_知乎茶也』测评",
60
+ "2018-02-05_科研二三事",
61
+ "2018-02-06_舆论常常是私刑 私刑常常是娱乐",
62
+ "2018-02-11_一撮热卤一撮湘",
63
+ "2018-02-13_如果你愿意一层一层地剥开这层茧",
64
+ "2018-02-17_衔瑜过年不被吃的办法",
65
+ "2018-03-01_只是不想取一个《深夜文摘》那样的题目",
66
+ "2018-03-14_下笔太狠的雨声 拈花杀了诗人",
67
+ "2018-04-02_斐波那契的日程表",
68
+ "2018-04-13_几个最近的小波动",
69
+ "2018-05-03_清醒与清醒的泥沼",
70
+ "2018-05-04_有道理&&没有价值",
71
+ "2018-05-18_拨乱反正赠送的语肝",
72
+ "2018-05-28_留下尬走的足迹_再到上海",
73
+ "2018-06-30_我是张衔瑜 我回来了啦",
74
+ "2018-07-08_明明说了话 别人却不明白",
75
+ "2018-07-10_无病呻吟",
76
+ "2018-07-16_我在北京 因为药神而多走了两步",
77
+ "2018-07-31_文不加点的张衔瑜有一岁了",
78
+ "2018-08-09_朴素思路的若干重载",
79
+ "2018-08-12_留下尬走的足迹_北京",
80
+ "2018-08-25_如果用脑内来看最近新闻中的问题,那我就有了两组问题",
81
+ "2018-09-05_你是一支聚酯,好久没有再回来过。",
82
+ "2018-09-12_湘江水逝楚云飞 _ 域见",
83
+ "2018-10-07_留下尬走的足迹_新加坡",
84
+ "2018-10-31_重拾《倚天屠龙记》后的随笔",
85
+ "2018-11-21_关于光谷我知道的不多",
86
+ "2018-11-30_鲁磨路有在过马路",
87
+ "2018-12-02_和郁结坐下来谈妥一次",
88
+ "2018-12-12_汉阳火车站_一枚武汉的古旧印章",
89
+ "2018-12-21_我今作力所能及的解说",
90
+ "2018-12-22_修正秩序当中那些不万能的部分",
91
+ "2018-12-24_恋爱要不要苟延残喘?",
92
+ "2018-12-31_总是容易被往事打动_100图和衔瑜的2018再见",
93
+ "2019-01-06_失焦的注意力",
94
+ "2019-01-12_我吃外卖披萨的时候一定要看成龙历险记(点头",
95
+ "2019-01-15_未写完的不具名",
96
+ "2019-01-20_留下尬走的足迹_长春",
97
+ "2019-01-26_留下尬走的足迹_合肥",
98
+ "2019-02-02_长沙人是检验臭豆腐的唯一标准",
99
+ "2019-02-07_为什么还要继续做现在在做的事呢",
100
+ "2019-02-11_当代戏精文学与沙雕图研究",
101
+ "2019-02-24_戏,是这样多吗?",
102
+ "2019-03-04_追着光梦游_一周图集",
103
+ "2019-03-25_生活应该怎样继续",
104
+ "2019-03-27_“对抗焦虑”是如何被制造出来的?",
105
+ "2019-03-29_别说跳出舒适圈,现在的问题是没有舒适圈",
106
+ "2019-03-31_如果没有这碗肠粉,今天不过又是个普通的周末",
107
+ "2019-04-03_私奔,在武昌汉口的生活里",
108
+ "2019-04-15_留下尬走的足迹_深圳",
109
+ "2019-04-21_暴戾的标准化流程",
110
+ "2019-04-23_报复性熬夜·暴食·随笔",
111
+ "2019-04-28_命运这辆公交车,我没有座位",
112
+ "2019-05-01_衔瑜,这里补交七篇读后感",
113
+ "2019-05-12_连续加班十九天之后",
114
+ "2019-05-13_第十三周周一的早课和白日梦",
115
+ "2019-05-16_豆浆和摘抄本",
116
+ "2019-05-18_紫菘停水停电停网前后24小时",
117
+ "2019-05-23_我们重视坎本身,更甚于坎的意义",
118
+ "2019-05-28_一点迷惑行为观察",
119
+ "2019-05-31_送毕业后的碎碎念",
120
+ "2019-06-06_朝花夕拾,组图扫街",
121
+ "2019-06-16_作为过来人,我告诉你 别过来",
122
+ "2019-06-17_末代皇帝_影评",
123
+ "2019-06-22_雨季,转凉的下午,等不到放晴",
124
+ "2019-06-25_能够正常安分地过完,就已经是个奇迹了",
125
+ "2019-06-29_打翻可乐后又蹭倒了椅子",
126
+ "2019-07-07_疲劳驾驶一千两百公里",
127
+ "2019-07-09_小张小张,从不慌张",
128
+ "2019-07-15_小张小张,卑微开张",
129
+ "2019-07-19_凌晨一点、卸下伪装、露出素颜,称为“丑时”",
130
+ "2019-07-21_青春万岁,都要颗粒归仓",
131
+ "2019-07-24_空腹的人可以吃饭吗?",
132
+ "2019-07-29_一只沙雕叼走了一颗栗子",
133
+ "2019-07-31_文不加点的张衔瑜两岁了",
134
+ "2019-08-18_与恶,只有一点点距离",
135
+ "2019-08-26_留下尬走的足迹_香港、深圳",
136
+ "2019-08-30_唱片店回忆札",
137
+ "2019-09-14_留下尬走的足迹_厦门",
138
+ "2019-09-14_留下尬走的足迹_皖赣边界",
139
+ "2019-09-22_留下尬走的足迹_兰州",
140
+ "2019-09-26_一天过完,绕着远路回去",
141
+ "2019-09-28_过气迷惑战士",
142
+ "2019-10-09_为什么要怕鬼?害你的全是人。",
143
+ "2019-10-12_我吃饭的时候不饿",
144
+ "2019-10-21_路灯底下,没有我要找的钥匙",
145
+ "2019-10-28_今天犯点什么坏呢?",
146
+ "2019-11-03_没必要抱有希望,没必要不抱希望",
147
+ "2019-11-09_攒够了失望",
148
+ "2019-11-14_重症自闭:成为自己的NPC",
149
+ "2019-11-15_穿着衬衫睡觉",
150
+ "2019-11-23_冬郁症",
151
+ "2019-12-06_我饿了。看来漂亮不能当饭吃。",
152
+ "2019-12-27_留下尬走的足迹_襄阳",
153
+ "2019-12-28_恶性社会事件,不必变相“荡妇羞辱”自己",
154
+ "2020-01-29_自由二字,长得条条框框",
155
+ "2020-02-04_把生活的书页翻得哗啦啦地响",
156
+ "2020-02-10_人不堪其忧,我也不改其乐",
157
+ "2020-03-02_情绪过载",
158
+ "2020-04-11_我的人文启蒙",
159
+ "2020-04-29_尝试用一句话解释复杂生活",
160
+ "2020-06-13_像复仇一样出门玩",
161
+ "2020-07-18_事件持续发酵,最后松软可口",
162
+ "2020-07-31_文不加点的张衔瑜三岁了",
163
+ "2020-08-15_日落会在夏天开放吗",
164
+ "2020-08-18_钝感隔离",
165
+ "2020-09-03_一只壁虎钻进鞋柜不见",
166
+ "2020-09-12_摸鱼,是我最想传承的非物质文化遗产",
167
+ "2020-09-28_喜欢一种风格就拿相机去拍到吐",
168
+ "2020-10-05_因为春天很闲,秋天很累",
169
+ "2020-10-31_生 活 档 案 盒",
170
+ "2020-11-10_人山人海,边走边忘",
171
+ "2020-11-17_印裔的新年:一场表演力的狂欢",
172
+ "2020-11-20_和任何一种生活摩擦久了,都会起球",
173
+ "2020-11-26_在机场等一艘船来",
174
+ "2020-12-01_导航:请在合适位置掉头",
175
+ "2020-12-10_地铁看书,坐过了要下的站 _ 2020年书单",
176
+ "2020-12-16_这哪是回笼觉?这是猪笼觉。",
177
+ "2020-12-19_生是我的五花肉,熟是我的回锅肉",
178
+ "2020-12-30_你说别追呀又依依不舍_100图和衔瑜的2020再见",
179
+ "2021-01-17_醒,在想等会儿梦什么",
180
+ "2021-01-19_一杯橘子汽水就能灌满外耳道",
181
+ "2021-02-05_好久没吃杨枝甘露",
182
+ "2021-02-10_一杯Kopi C Kosong也还是困",
183
+ "2021-02-14_在坡的辛丑新年",
184
+ "2021-03-05_一杯过饱和又不断析出的溶液",
185
+ "2021-03-15_你还(huán)什么咯?",
186
+ "2021-03-18_盐王爷",
187
+ "2021-03-26_有了上顿盼下顿",
188
+ "2021-03-30_智商散装[图]出入",
189
+ "2021-04-03_电梯门开",
190
+ "2021-04-13_回车键诗人",
191
+ "2021-04-29_晒得人灭涝的",
192
+ "2021-05-25_忘性大的人不会老",
193
+ "2021-06-17_留下尬走的足迹_再到北京",
194
+ "2021-07-16_尽量少用偏正短语",
195
+ "2021-07-24_留下尬走的足迹_天津北京武汉",
196
+ "2021-08-04_听君一席话,白听一席话",
197
+ "2021-08-21_写段子是一回事",
198
+ "2021-09-11_留下尬走的足迹_桂林",
199
+ "2021-09-12_耳朵占线",
200
+ "2021-09-20_没头没尾",
201
+ "2021-10-02_留下尬走的足迹_敦煌",
202
+ "2021-10-17_哪能什么细节都严丝密缝",
203
+ "2021-11-02_留下尬走的足迹_北海涠洲南宁上海",
204
+ "2021-11-08_留下尬走的足迹_梅州汕头",
205
+ "2021-11-16_气温每降低一度,炒板栗店就多一队人",
206
+ "2021-11-24_留下尬走的足迹_宁波义乌杭州贵阳",
207
+ "2021-12-07_留下尬走的足迹_南京镇江扬州盐城南通",
208
+ "2021-12-27_留下尬走的足迹_山西北段中段",
209
+ "2021-12-30_梦把我从万花筒坠落_100图和衔瑜的2021再见",
210
+ "2022-01-10_我发现实际上来说今天开学了",
211
+ "2022-01-20_严丝密缝仍是最大问题",
212
+ "2022-02-08_什么可以是一味安定?书店吗",
213
+ "2022-02-12_我也会自转和公转",
214
+ "2022-02-19_闲聊波尔卡",
215
+ "2022-03-06_光彩奇异的三月生活",
216
+ "2022-03-17_焦虑突破了血脑屏障",
217
+ "2022-04-08_你说什么叫路径依赖",
218
+ "2022-04-20_留下尬走的足迹_湘西",
219
+ "2022-05-08_拼凑的碎片不能叫整体",
220
+ "2022-05-20_没有遗憾了才可以睡觉",
221
+ "2022-05-27_道拉基",
222
+ "2022-06-12_有人路过那里",
223
+ "2022-06-26_消失在人海",
224
+ "2022-07-03_原来如实",
225
+ "2022-07-31_文不加点的张衔瑜五岁了",
226
+ "2022-08-19_留下尬走的足迹_第很多次到北京",
227
+ "2022-09-09_严禁盗采盗伐",
228
+ "2022-09-18_在公园长椅睡觉才是正经事",
229
+ "2022-09-21_这个天气骑车还有点冷",
230
+ "2022-11-25_悲伤蛙出镜好多次",
231
+ "2022-12-02_眼神聚焦是一件难事",
232
+ "2022-12-22_奇怪的人生经历增加了",
233
+ "2022-12-29_升降机送上一桶水泥_100图和衔瑜的2022再见",
234
+ "2023-01-24_连日降水",
235
+ "2023-02-10_தைப்பூசம்大宝森节行记(微恐慎入",
236
+ "2023-02-19_一行文字要读两遍",
237
+ "2023-02-26_构思一个身份、然后出门",
238
+ "2023-03-17_闭着眼睛坐公交车",
239
+ "2023-04-09_留下尬走的足迹_新山",
240
+ "2023-04-17_朋友说看见炸烤串的很像我",
241
+ "2023-04-29_搞抽象",
242
+ "2023-04-30_再搞抽象",
243
+ "2023-05-07_消遣",
244
+ "2023-05-11_罗生门之抽象",
245
+ "2023-05-21_和农忙过年一样忙",
246
+ "2023-06-07_留下尬走的足迹_福州",
247
+ "2023-06-08_留下尬走的足迹_苏州上海",
248
+ "2023-06-14_被人说我干活时讲话就像英翻中",
249
+ "2023-07-06_萝卜快了不洗泥",
250
+ "2023-07-19_两部电影的观后小札",
251
+ "2023-08-13_鬼画联翩",
252
+ "2023-08-31_像一袋黄豆一样坐在那",
253
+ "2023-08-31_像一袋黄豆一样坐在那(GPT版)",
254
+ "2023-09-04_心狠点下老贵外卖",
255
+ "2023-09-14_先关电梯门,再按键",
256
+ "2023-10-06_但馄饨很烫",
257
+ "2023-10-11_没擦通风口",
258
+ "2023-11-06_逛松弛感裤子越逛越紧张",
259
+ "2023-11-06_风能不能凉一点",
260
+ "2023-11-19_夸大风险评估",
261
+ "2023-11-20_盯着鞋子",
262
+ "2023-12-12_眾鳥非鳥",
263
+ "2023-12-21_树莓粉和捣练图",
264
+ "2023-12-29_只要内心远过空旷_100图和衔瑜的2023再见",
265
+ "2024-01-05_荒腔走板",
266
+ "2024-01-17_度百天,然后等夜晚",
267
+ "2024-01-29_盘盘焉,囷囷焉",
268
+ "2024-02-06_一排路灯坏了好几个整流器",
269
+ "2024-02-24_胡说八道的乐趣",
270
+ "2024-04-07_连接",
271
+ "2024-04-16_器物感",
272
+ "2024-04-22_下班后我就不戴眼镜了",
273
+ "2024-05-01_脚底踩了个鞭炮",
274
+ "2024-05-06_红顶棚灰铁架",
275
+ "2024-05-17_乱成一锅粥,喝粥",
276
+ "2024-06-09_不打转向灯",
277
+ "2024-06-13_戴降噪耳机听白噪音",
278
+ "2024-06-17_打胡须膏的时候",
279
+ "2024-06-23_船头跟我点点头",
280
+ "2024-06-27_趔趄",
281
+ "2024-06-30_赌",
282
+ "2024-07-05_忘",
283
+ "2024-07-08_梦",
284
+ "2024-07-13_牛羊也下山了",
285
+ "2024-07-17_悟已往之不谏",
286
+ "2024-07-22_看电影",
287
+ "2024-07-28_梦到世纪大停电",
288
+ "2024-07-31_文不加点的张衔瑜七岁了",
289
+ "2024-08-22_留下尬走的足迹|西马、新加坡",
290
+ "2024-09-03_留下尬走的足迹|大湾区",
291
+ "2024-09-13_很精彩",
292
+ "2024-09-22_去了一趟月球",
293
+ "2024-09-26_在庙里住了一天",
294
+ "2024-09-30_太阳无非一颗晨星",
295
+ "2024-10-09_火柴盒",
296
+ "2024-10-17_溨雨",
297
+ "2024-10-29_留下尬走的足迹|京津",
298
+ "2024-11-02_则其负大舟也无力",
299
+ "2024-11-05_一本糊涂账",
300
+ "2024-11-13_备忘录空空",
301
+ "2024-11-27_留下尬走的足迹|巴淡岛",
302
+ "2024-12-24_留下尬走的足迹|巴厘岛爪哇岛",
303
+ "2024-12-25_留下尬走的足迹|曼谷、大城府",
304
+ "2024-12-26_留下尬走的足迹|泉州",
305
+ "2024-12-31_你更喜欢片面地醒着_100图和衔瑜的2024再见",
306
+ "2025-01-18_躺在与这个社会不痛不痒",
307
+ "2025-01-28_听风",
308
+ "2025-02-09_高功耗",
309
+ "2025-02-16_晒太阳是被太阳晒",
310
+ "2025-02-23_大气层没有居民",
311
+ "2025-03-03_认知武器",
312
+ "2025-03-10_出入是门",
313
+ "2025-03-18_变化快",
314
+ "2025-03-24_小雷音",
315
+ "2025-04-01_热了冷,冷了热",
316
+ "2025-04-07_明亮",
317
+ "2025-04-14_预制菜",
318
+ "2025-04-20_检票员",
319
+ "2025-04-28_重在掺和",
320
+ "2025-05-08_状况外",
321
+ "2025-05-15_留下尬走的足迹|昆明楚雄遵义三峡",
322
+ "2025-05-26_放空",
323
+ "2025-06-02_量体裁衣",
324
+ "2025-06-10_吃禽不是嘴过",
325
+ "2025-06-17_让位",
326
+ "2025-06-24_语言光滑",
327
+ "2025-07-06_這下好了,是這下壞了",
328
+ "2025-07-14_垂",
329
+ "2025-07-22_在天垂形",
330
+ "2025-07-28_Perplexity 困惑度",
331
+ "2025-07-31_文不加点的���衔瑜八岁了",
332
+ "2025-08-09_自来水泵",
333
+ "2025-08-18_夏天热冬天冷",
334
+ "2025-08-27_第八十万字",
335
+ "2025-09-10_溶解",
336
+ "其他"
337
+ ],
338
+ "sample_templates": [
339
+ "改写",
340
+ "续写",
341
+ "总结展开"
342
+ ],
343
+ "config_used": {
344
+ "filter_years": null,
345
+ "train_output": "data/sft_train.jsonl",
346
+ "val_output": "data/sft_val.jsonl",
347
+ "stats_output": "data/dataset_stats.json",
348
+ "val_split": 0.1,
349
+ "random_seed": 42,
350
+ "min_chunk_length": 300,
351
+ "max_chunk_length": 800,
352
+ "min_article_length": 100,
353
+ "enable_rewrite_template": true,
354
+ "enable_continue_template": true,
355
+ "enable_summarize_template": true,
356
+ "enable_qa_template": false,
357
+ "enable_deduplication": true,
358
+ "similarity_threshold": 0.8,
359
+ "enable_advanced_dedup": true,
360
+ "verbose": true,
361
+ "max_articles": null,
362
+ "enable_quality_filter": true,
363
+ "min_unique_chars": 10,
364
+ "min_sentence_count": 1
365
+ },
366
+ "deduplication_enabled": true,
367
+ "advanced_dedup_enabled": true,
368
+ "quality_filter_enabled": true
369
+ }
data/sft_train.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
data/sft_val.jsonl ADDED
The diff for this file is too large to render. See raw diff