zhaode/EagleChat

EagleChat是一个高质量、经过精心整合的中英双语对话指令微调数据集。该数据集的核心目标是为大语言模型提供一个能够显著提升其综合对话能力的优质语料。通过融合ShareGPT、UltraChat 200k和smoltalk-chinese三个数据集的优点，内容丰富多样，并包含大量高质量的中文和英文对话，有助于提升模型的跨语言能力。数据集已在EAGLE模型上验证过效果，能提升模型的对话流畅性、指令遵循能力和综合表现，且数据已清洗和格式化，可直接用于主流的微调框架。

EagleChat is a high-quality, meticulously curated bilingual (Chinese & English) conversational dataset for instruction fine-tuning. The primary goal of this dataset is to serve as a premium corpus to significantly enhance the comprehensive conversational abilities of Large Language Models, especially models like EAGLE. It combines the strengths of three datasets: ShareGPT, UltraChat 200k, and smoltalk-chinese, offering rich and diverse content. The dataset includes a large number of high-quality Chinese and English conversations, which helps to improve the cross-lingual capabilities of the model. It has been empirically proven that fine-tuning the EAGLE model with EagleChat leads to significant performance improvements in terms of conversational smoothness, instruction following ability, and overall performance. The data is cleaned and formatted, ready to be used in mainstream fine-tuning frameworks.