Truthseeker87/solarhive-community-solar-1k

SolarHive社区太阳能1K数据集包含1,029个训练示例，用于社区太阳能能源智能领域，涵盖9个能源领域、15+美国城市和4个实时API数据源。数据集分为两部分：413个手工制作的静态示例和516个基于API生成的示例。静态示例覆盖15+美国城市和9个能源领域，包含专家编写的问答，涵盖边缘案例、紧急情况和多步推理。API生成的示例包括501个基于实时API数据的问答和100个工具调用示例。数据集还包含工具调用格式的示例，用于训练Gemma 4模型的功能调用能力。数据质量通过12个诊断图表进行验证，确保所有数字声明都追溯到真实的API响应。数据集的使用方法和相关资源也在README中进行了详细说明。The SolarHive Community Solar 1K Dataset contains 1,029 training examples for community solar energy intelligence, covering 9 energy domains, 15+ US cities, and 4 live API data sources. The dataset is divided into two parts: 413 hand-crafted static examples and 516 API-grounded generated examples. The static examples provide breadth across 15+ US cities and 9 energy domains, with expert-written Q&A covering edge cases, emergencies, and multi-step reasoning. The API-generated examples include 501 Q&A generated from live API data and 100 tool-calling examples. The dataset also includes examples in a tool-calling format to train Gemma 4s native function calling capabilities. Data quality is validated through 12 diagnostic charts, ensuring all numeric claims trace back to real API responses. The README also provides detailed instructions on how to use the dataset and related resources.