iapp/dpo_thai_tutorial
收藏Hugging Face2026-02-04 更新2026-02-07 收录
下载链接：
https://hf-mirror.com/datasets/iapp/dpo_thai_tutorial
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是为学习和演示Direct Preference Optimization (DPO)技术而创建的泰语数据集。包含100个真实例子，涵盖多个主题（金融、经济、常识等）。每个例子包含泰语问题、原始回答、优质泰语回答（chosen）和低质量回答（rejected，通常是英语或不相关回答）。数据集主要用于DPO学习、语言对齐和质量对齐。
This dataset is created for studying and demonstrating DPO techniques for LLM alignment. It contains 100 real-world examples covering various topics (finance, economics, general knowledge). Each example includes a Thai question, original answer, high-quality Thai answer (chosen), and low-quality answer (rejected, often in English or irrelevant). The dataset is primarily used for DPO learning, language alignment, and quality alignment.
提供机构：
iapp
