maya-research/IndicVault

Indic Vault是一个面向聊天机器人和语音助手的高质量、指令微调的数据集，包含2025年印度日常使用的当代语言中的问题回答对。它涵盖了20个必要主题，包括金融、健康、技术、人际关系等，以自然、口语化的语言风格为特点，提供有意义且上下文相关的信息。该数据集支持从4B到27B参数的开源模型进行微调。

Indic Vault is a high-quality, instruction-tuned dataset featuring question-answer pairs in the contemporary, everyday language spoken across India in 2025. It covers 20 essential topics such as Finance, Health, Technology, Relationships, and more, characterized by natural, colloquial language and meaningful, contextually relevant information. The dataset supports fine-tuning on open-source models ranging from 4B to 27B parameters.