Table 3. Summary of the real-world datasets.

Dataset	Dimension $d$	Distance	Index size $n$	Query size $n\mathrm{\prime }$	Related reference
F-MNIST	784	Euclidean	25,000	10,000	Xiao, Rasul & Vollgraf (2017)
SIFT10K	128	Euclidean	25,000	100	Lowe (2004)
SIFT1M	128	Euclidean	100,000	10,000	Lowe (2004)
GIST	960	Euclidean	1,000,000	1,000	Oliva & Torralba (2004)
GloVe100	100	Angular	1,183,514	10,000	Pennington, Socher & Manning (2014)
DEEP1B	96	Angular	9,990,000	10,000	Yandex & Lempitsky (2016)