RoomR

该数据集名为RoomR，包含了训练集、验证集和测试集三个部分。训练集包含80个房间和4000个任务，验证集包含20个房间和1000个任务，测试集同样包含20个房间和1000个任务，旨在评估代理在将房间重新布置成期望物体目标配置方面的能力。此外，该数据集还包含了一个两阶段的设置，代理首先观察场景的目标配置，然后在物体被随机打乱后，将物体重新排列回其目标配置。规模上，数据集涵盖了80个训练房间、20个验证房间以及20个测试房间。任务方面，主要是视觉房间重新排列。

This dataset is named RoomR, which comprises three subsets: the training set, validation set, and test set. The training set includes 80 rooms and 4000 tasks, the validation set contains 20 rooms and 1000 tasks, and the test set also has 20 rooms and 1000 tasks. It is designed to evaluate the capability of AI Agents in rearranging rooms to match the desired target object configurations. Furthermore, this dataset incorporates a two-stage experimental setup: the agent first perceives the target configuration of the scene, then rearranges the randomly disturbed objects back to their target positions. In terms of scale, the dataset covers 80 training rooms, 20 validation rooms, and 20 test rooms. The core tasks of this dataset focus on visual room rearrangement.