G-A-I/GraphOmni
收藏Hugging Face2025-05-28 更新2025-11-01 收录
下载链接：
https://hf-mirror.com/datasets/G-A-I/GraphOmni
下载链接
链接失效反馈官方服务：
资源简介：
GraphOmni数据集为评估大型语言模型在图问题上的推理能力提供了一个统一的工作流程。它支持六种图算法任务，包括连通性、广度优先搜索顺序、三角形计数、直径、环检测和最短路径。该数据集包含超过24万个示例，适用于各种模型、图表示、提示类型和任务制定，可以帮助自动评估和跟踪不同提示和建模设置的结果。
The GraphOmni dataset provides a unified pipeline to evaluate large language models (LLMs) on graph reasoning capabilities. It supports six graph algorithm tasks including Connectivity, Bfsorder, Triangle, Diameter, Cycle, and Shortest path. The dataset contains over 240,000 examples, applicable to various models, graph representations, prompt types, and task formulations, enabling automated evaluation and result tracking across different prompting and modeling setups.
提供机构：
G-A-I
