twinkle-ai/finetranslations-zhtw

本数据集旨在构建一套高质量的多语言到繁体中文（zh-tw）机器翻译数据集，扩展Hugging Face生态系统中已有的多语言到英文翻译数据集。通过结合大型语言模型（LLM）的初步翻译和具备源语言专业能力的人工审核，数据集致力于提升翻译质量，支持以繁体中文为核心的多语言机器翻译和跨语言NLP研究，改善低资源语言配对的翻译稳定性，并提供可扩展、可重现的数据集构建流程。

This dataset aims to build a high-quality multilingual to Traditional Chinese (zh-tw) machine translation dataset, extending the existing multilingual to English translation dataset in the Hugging Face ecosystem. By combining preliminary translations from large language models (LLMs) with human review by professionals with source language expertise, the dataset strives to improve translation quality, support Traditional Chinese-centric multilingual machine translation and cross-lingual NLP research, enhance translation stability for low-resource language pairs, and provide a scalable, reproducible dataset construction process.