Turbo
data class Turbo(val bits: TurboQuantBitSize? = null, val alwaysRam: Boolean? = null, val memory: Memory? = null) : QuantizationConfig
TurboQuant, at one of four bit sizes.
null bits leaves the server's default. This is quantization over full-precision storage; to store nothing but 4-bit vectors and keep no originals, set VectorDatatype.TURBO4 on the vector instead, which is a different decision about what the collection holds.