benchang1110/Chattw_v2

---
dataset_info:
features:
- name: conversation
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: train
num_bytes: 180805392.13851646
num_examples: 164463
download_size: 110324568
dataset_size: 180805392.13851646
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

## Dataset summary
This is a synthetic dataset. This dataset uses the first 40% of the prompt from [yentinglin/TaiwanChat](https://huggingface.co/datasets/yentinglin/TaiwanChat).
The reply from the assistant is generated by LLM. The dataset have been filtered to avoid repitition of the content generated by LLM.

## Future work
Finish the last 60% of [yentinglin/TaiwanChat](https://huggingface.co/datasets/yentinglin/TaiwanChat).

原始信息汇总

数据集概述

数据集特征

• conversation
  • content: 数据类型为字符串
  • role: 数据类型为字符串

数据集分割

• train
  • num_bytes: 180805392.13851646
  • num_examples: 164463

数据集大小

• download_size: 110324568
• dataset_size: 180805392.13851646

数据集来源与处理

• 本数据集为合成数据，使用了yentinglin/TaiwanChat数据集的前40%作为提示。
• 助手的回复由大型语言模型（LLM）生成，并已过滤以避免内容重复。

未来工作

• 完成yentinglin/TaiwanChat数据集剩余的60%。