LocationReasoner
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/miho-koda/LocationReasoner
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是一个旨在评估大型语言模型在实际场地选址任务中的推理能力的基准测试，包含了超过300个精心设计、难度各异的查询。此外，该基准测试还提供了一个沙盒环境，其中包含了基于约束的地点搜索工具，并评估了不同人工智能模型在这些任务上的表现。规模上，数据集包含了超过300个查询，任务重点在于场地选址的推理能力。
This dataset is a benchmark aimed at evaluating the reasoning capabilities of large language models (LLMs) in real-world site selection tasks. It includes over 300 meticulously designed queries with varying levels of difficulty. Furthermore, the benchmark offers a sandbox environment equipped with constraint-based site search tools, and assesses the performance of different artificial intelligence models on these tasks.
提供机构：
In-house
