Vector Compression for StaticMemoryIndex or DynamicMemoryIndex?

[Kunal Dahiya]

Hi

I wanted to check in regarding compression/quantization for in-memory index. From what I understand, it supports PQ (using 'use_pq_build') during graph construction, however, in the end, it keeps the full vectors in memory. Is there any option to keep compressed vectors in memory? Any other information regarding saving on memory for in-memory index is also welcome.

PS: This option is already supported in StaticDiskIndex via pq_disk_bytes.