ruhulu/TravelPlanner

TravelPlanner是一个专为评估语言代理在工具使用和多约束复杂规划中的能力而设计的基准数据集。该数据集包含1,225个查询，旨在测试代理在复杂规划的广度和深度上的能力。每个查询都要求语言代理制定一个全面的旅行计划，包括交通、每日餐饮、景点和住宿。数据集分为训练集（45个查询-计划对）、验证集（180个查询）和测试集（1,000个查询），每个查询都包含详细的记录布局，如出发城市、目的地城市、旅行天数、访问城市数量、日期、人数、本地约束、查询、难度级别、人工注释的计划和参考信息。

TravelPlanner is a benchmark crafted for evaluating language agents in tool-use and complex planning within multiple constraints. The dataset comprises 1,225 queries designed to test agents abilities across both the breadth and depth of complex planning. For each query, language agents are expected to formulate a comprehensive plan that includes transportation, daily meals, attractions, and accommodation for each day. The dataset is split into a training set (45 query-plan pairs), a validation set (180 queries), and a test set (1,000 queries), with each query including detailed record layouts such as origin city, destination city, number of days, number of cities visited, date, number of people, local constraints, query, difficulty level, human-annotated plan, and reference information.