wangzh12023/HuChenFeng-live-dialogue

HuChenFeng Live Dialogue Dataset

Dataset Introduction

This dataset contains dialogue data from Hu Chenfeng's live streaming room, suitable for tasks such as Chinese dialogue generation and chatbot training.

Dataset Statistics

• —Total Samples: \~200K dialogue turns
• —Language: Chinese
• —Format: Parquet, JSONL
• —License: MIT License

Data Format

Each data sample includes a messages field, which stores the dialogue history:

{
    "messages": [
        {
            "role": "user",
            "content": "就是说，你说中医完全没有作用，那它能作为一种宣传吗？中国文化的宣传。"
        },
        {
            "role": "assistant",
            "content": "你都说了作为文化是可以的，没问题啊，作为文化是可以的。"
        }
    ]
}

Field Descriptions

• —messages: List of dialogue messages
• —role: Role type, user or assistant
• —content: Message content

Usage

Quick Loading

from datasets import load_dataset

# Method 1: Load Parquet format (Recommended, faster)
dataset = load_dataset("wangzh12023/HuChenFeng-live-dialogue")

# Method 2: Load JSONL format
dataset = load_dataset(
    "json", 
    data_files="hf://datasets/wangzh12023/HuChenFeng-live-dialogue/train.jsonl"
)

Data Source

The data is sourced from public dialogue records in Hu Chenfeng's live streaming room, refer, processed through cleaning and formatting.

Data Quality

• —✅ Empty messages filtered
• —✅ Complete dialogue context preserved
• —✅ Unified message format
• —✅ Primarily Chinese language

Applicable Scenarios

• —🤖 Chinese chatbot training
• —💬 Dialogue generation model fine-tuning
• —📚 Natural Language Processing research
• —🎓 Educational and learning purposes

Citation

If you use this dataset, please cite:

@dataset{huchenfeng_live_dialogue_2025,
  title={HuChenFeng Live Dialogue Dataset},
  author={Zihan Wang},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/wangzh12023/HuChenFeng-live-dialogue}}
}

Contact Information

• —Dataset Maintainer: wangzh12023
• —Hugging Face: @wangzh12023

Update Log

v1.0.0 (2025-10-20)

• —Initial release
• —Contains \~200K dialogue data samples
• —Provides both Parquet and JSONL formats

v1.1.0 (2025-10-21)

• —Add system role in messages lists

v1.2.0 (2025-10-21)

• —Reconstruct json, regenerate datasets ---

Note: This dataset is for research and educational purposes only. Please comply with relevant laws and regulations.

Acknowledgements

We acknowledge the open-source community for providing standard practices. The Markdown template and general structure for this dataset card were based on a this public repository