AndyMorgan-NOIRLab/BGRS

BGRS数据集包含了来自FITS HDU 1的标量元数据和从HDF5中提取的每行的密集向量。该数据集总共包含1089977行数据，经过筛选后，训练和测试集合并共有1083363行数据。向量维度为7781。数据集分为三个部分：全部数据（full）、训练集（train）和测试集（test）。训练集是经过筛选的80%，而测试集是20%。筛选标准为GALAXY的delta_z小于0.0033或QSO的delta_z小于0.01。The BGRS dataset contains scalar metadata from FITS HDU 1 and dense per-row vectors extracted from HDF5. The dataset consists of a total of 1089977 rows, with 1083363 rows after filtered selection for train and test combined. The vector dimension is 7781. The dataset is split into three parts: the full dataset, the training set (train), and the test set (test). The training set is 80% of the filtered selection, while the test set is 20% of it, with the selection criterion being GALAXY delta_z < 0.0033 or QSO delta_z < 0.01.