Hugging Face's logo

Datasets:
benchang1110
/
Taiwan-pretrain-9B

Modalities:
Text
Formats:
parquet
Size:
1M - 10M
Libraries:
Datasets
Dask
Dataset card Viewer Files Community
1

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.

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

Dataset summary

This dataset is designed for Traditional Chinese (zh-tw) and comprises a collection of texts from various sources, including news articles, scientific publications, technological reports, and Wikipedia entries. The number of tokens from each source is listed below.

Wikipedia

Science

Technology

Total tokens: 9.1B

(Tokens are calculated by tokenizer of LLaMA2)

Usage 

from datasets import load_dataset

dataset = load_dataset("benchang1110/Taiwan-pretrain-9B", split="train")
Downloads last month
33

Collection including benchang1110/Taiwan-pretrain-9B