Datasets:
BAAI
/

Languages:
Chinese
ArXiv:
License:

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

You agree to not use the dataset to conduct experiments that cause harm to human subjects.

Log in or Sign Up to review the conditions and access this dataset content.

Data Description

To address the scarcity of high-quality safety datasets in the Chinese, we open-sourced the CCI (Chinese Corpora Internet) dataset on November 29, 2023. Building on this foundation, we continue to expand the data source, adopt stricter data cleaning methods, and complete the construction of the CCI 3.0 dataset. This dataset is composed of high-quality, reliable Internet data from trusted sources. It has undergone strict data cleaning and de-duplication, with targeted detection and filtering carried out for content quality and safety. The rules for data processing include:

  • Rule-based filtering: safety filtering based on keywords, spam information filtering, etc.
  • Model-based filtering: filtering of low-quality content by training a classification model
  • Deduplication: within and between datasets dedup

Besides, we added rich meta information including quality score and educational level tagged by small models. Users can conveniently utilize the meta information of each data entry to further filter and customize the dataset. The CCI 3.0 corpus released is about 1000GB in size.

Update

  • Oct 25, 2024, CCI 3.0 HQ Tech Report released!
  • Sep 20, 2024, CCI 3.0 released!

Data Format

Field Type Meaning
id String Document ID, globally unique
content String Content of the document
meta_info String Meta Info of the document

Sample

{
  "id": "02301a3477ca2b5434ab29dfc32f95d853abc",
  "content": "《农村财政与财务》杂志创办于1996,是中国农村财政研究会主管的国家重点学术期刊,国家级期刊,影响因子0.163,现被万方收录(中)等权威机构收录,主要方向:研究报告、文献综述、简报、专题研究\n《农村财政与财务》以宣传党和国家财政政策、推动税收体制改革、研究财税理论、指导基层财政和涉农工作,传播理财知识为宗旨,融政策性、指导性、权威性、实用性和知识性为一体。\n《农村财政与财务》是贯彻国家方针、政策、探索财税理论和有关难点、热点问题,交流财政科学化、精细化管理经验,帮助读者提高综合素质和政策水平不可或缺的理想媒体。\n中共中央办公厅国务院办公厅印发《关于加快构建政策体系培育新型农业经营主体的意见》\n9月5号投的,15号就给了初审结果,给出的修改意见,主要是篇幅过长,以及图片格式的问题。修改后过了一周,就发录用通知了。皇天不负有心人啊,继续努力。\n两个意见,总体来看属于一个大修,一个小修,编辑要求修改后复审。但是意见真的给的很中肯,用了一个星期时间认真修改。提交修改稿后,编辑部很快送出外审,当天外审专家就完成了复审工作,然后在第二天立马显示接收了。这个复审速度吓得我惊人,不敢相信是被录用了,后来打电话确认已被录用,等待后续排版工作。\n两个审稿人,审理比较负责,给出了几点小建议,属于小修,修改后录用,编辑对全文进行了细致标注,对格式要求、图表制作规范较为严格,杂志效率挺高,尤其是编辑部反应神速,必须赞一个。\n农村财政与财务杂志的编辑和审稿人都非常专业,两个审稿人分别提出了3条和5条审稿意见,而且有些意见颇有意义,但是对我的文章还是非常肯定的,不到一个月消息回复审稿人分别要求大修和小修,要求比较严谨,数据比较足够,就能中。祝好运。\n农村财政与财务杂志速度还是很快的,而且是我见过的回复字数最多最多的编辑信,投稿一个月,反馈结果。修改后,递交编辑部,审稿人很心细,改的很认真。连标点居然都帮我改……修改两次后录用。\n编辑的工作十分点赞,态度也是很友善,审稿专家也是非常专业,虽然历经的时间比较长才录用,但是也情有可原,毕竟投稿量太大,而且期间加上放假,难免时间较长,进入编辑加工阶段后才进行了咨询,编辑也进行了详细的回复,希望对各位投稿有所帮助。\n农村财政与财务杂志编辑很负责,整个投稿流程节奏非常快。个人感觉这个杂志还是不错的。2位审稿人都比较专业,有个审稿人的一些意见还是非常有帮助,非常有针对性。速度也比较快。推荐大家投稿!\n第二年来订阅杂志了,客服的态度很好哦,杂志的寄送也还及时,希望以后对老顾客有一定的优惠。\n农村财政与财务杂志的审稿速度还是值得肯定的。综合来说,审稿人还是比较认真的,给修改的也比较仔细,对创新性要求还算比较高吧,编辑老师也非常的平易近人。虽然是第一次投稿,但是还是很幸运被收录了。个人建议文章比较注重自主创新,思维清晰。希望能对大家有帮助!\n农村财政与财务杂志效率很高的,也觉得自己蛮幸运的。当时看到外审两三天回来了,以为要被拒了呢,结果给修改意见了。两周后提交修改稿,两三天后显示录用了。整个下来小一个月吧,第一次投稿,还是感觉蛮幸运的。\n该刊审稿较快,出刊也快前后跨度就半年左右,编辑老师态度很好,最好使用邮箱投稿,外审一般会告知你,里面文章质量感觉都挺好的,良心杂志,介意普刊的同仁可以投投看!!\n农村财政与财务杂志质量不错,审稿较严格,录用较快。属于很规范的中文杂志。编辑很负责,处理也很快、工作规范,相当满意。审稿专家很认真细致,意见提的很详细,对论文提高很有帮助!相当愉快的一次投稿经历~\n总的来说,审稿专家还是蛮认真的,对待问题都很细致。另外,编辑也相当赞,经常打电话去咨询状态,一直很要是有创意,内容丰富,应该就没有问题。\neleme**:杂志工作人员的处理速度相当不错哦,审稿专家很负责。\nfazhi**:投稿后编辑态度不错,邮件联系均有及时回复。\n15年11月16日投稿,修改了两次,第一次对文章创新性提出了意见,第二次是格式方面的修改,12月15日通知正刊录用。算是比较快的了。该刊给人的第一感觉就是正规,对论文内容、格式等要求也很严格,应该认真对待。祝大家成功!\nxiajia**:很开心。总体来说,审稿速度很快,比较满意;可以试试。\n9月初投稿,一直没有消息,月底打电话问,还在外审。10月初收到退修通知,修改后返回,编辑回复很快,让修改了格式,然后通知录用。编辑很负责。等待校稿和版费通知。\njince**:感觉给出的意见很诚恳,很有建设性。\n初审大概一周左右,进入外审程序。8月底左右还是正在二审中,我打电话问了下,才告诉我需要修改,网上的状态变成“二审已审回”;按照修改意见修改后以电子邮件形式提交,大概一周后收到录用通知。\nsansui**:审稿速度还是相当神速,编辑部老师很好,很负责任。\n农村财政与财务速度蛮快的,编辑部也很负责,很有主见。审稿人信息反馈很快,20多天就有消息了,录用消息也第一时间通知,很及时、速度、高效,一点也不耽误时间。\n编辑非常认真负责,邮件联系回复也非常快,稿件开始本来有些问题,考虑不用的,但是编辑又给了一次修改的机会,说是修改好了还可能录用,就花心思修,修改后一个月不到就说录用了,还有一些小问题后面陆续解决了。\n用了两个月的时候,才被录用。审稿周期不短,可能也是自己写的不好一再返修的原因。觉得审稿人给的身高意见比较细致、对问题的提出比较准确。农村财政与财务的档次也很高。写的有点多所以相对的版面费也就要多一些。\nsusu**:个人感觉该期刊对文章的选题热点、创新点、写作水平都比较注重。\n个人感觉还不错。第一篇中的论文,还是很开心的。5月28号投稿7月15号通知录用。修改意见中,只有文中的格式问题以及图标中的,字体,单位问题。修改后就成功录用啦。\n农村财政与财务杂志的审稿速度飞快,貌似一个月左右就拟录用了,然后改了两次格式,缩小篇幅,大概也就一个半月搞掂。编辑部人员服务态度很好!很有耐心!大家可以尝试下这个杂志。",
  "url": "",
  "meta_info":
  {
    "safety_score": "0.0029210448265075684",
    "quality_score": 0.9920259714126587,
    "gzip_compressibility": 0.4772631873513362,
    "count_newlines": 31,
    "words_count": 1545,
    "stopword_ratio": 0.4181229773462783,
    "actionable_verb_count": 326,
    "actionable_verb_ratio": 0.21100323624595468,
    "2gram_repetition_ratio": 0.1016839378238342,
    "3gram_repetition_ratio": 0.0304601425793908,
    "entity_count": 424,
    "entity_dependency_count": 184
  }
}

Download

The CCI 3.0 dataset is simultaneously open-sourced on the BAAI DataHub and Huggingface.

BAAI DataHub

Users can click the link CCI 3.0 Dataset to view the data files, and click to download.

Note that users need to register on BAAI DataHub to use the data, and filling out a survey questionnaire is required before their first download.

Huggingface

To use the data, you can load it using the following code:

from datasets import load_dataset

dataset = load_dataset("BAAI/CCI3-Data")

Evaluation

Setup

Due to the mixed Chinese and English datasets, we chose Qwen2-0.5B model for datasets evaluation, each experiment with 100B tokens training.

We follow the same evaluation setup for all models using FineWeb setup with lighteval library. You can checkout the evaluation script here.

Results

We conducted two types of experiments:

  1. Mixed Dataset Experiment: The ratio of English, code, and Chinese is 60% : 10% : 30%.
  2. Chinese Dataset Experiment: The Chinese ratio is 100%.

For English datasets, we uniformly used FineWeb-edu. For code data, we used StarCoder. For Chinese datasets, we selected wanjuan-v1, skypile, and cci3.0.

In the plots below, cci_edu datasets is from CCI 3.0 HQ Dataset.

For Mixed Dataset Experiment all evaluation metrics are averaged and for Chinese Dataset Experiment only chinese evaluation metrics are averaged. Evaluation Metrics

All evaluation metrics across training are depicted in Evaluation Metrics Across Training.

Citation Information

You can cite our paper or this dataset:

@misc{wang2024cci30hqlargescalechinesedataset,
      title={CCI3.0-HQ: a large-scale Chinese dataset of high quality designed for pre-training large language models}, 
      author={Liangdong Wang and Bo-Wen Zhang and Chengwei Wu and Hanyu Zhao and Xiaofeng Shi and Shuhao Gu and Jijie Li and Quanyue Ma and TengFei Pan and Guang Liu},
      year={2024},
      eprint={2410.18505},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2410.18505}, 
}

User Agreement

Users need to comply with the usage agreement of the CCI 3.0 dataset. You can view the agreement by clicking on the following link: (View Usage Agreement).

Notice

If you have any questions related to this dataset, please contact [email protected].

Downloads last month
8,286

Models trained or fine-tuned on BAAI/CCI3-Data

Collection including BAAI/CCI3-Data