Add quantization capabilities for all models.

[EricApgar]

All models have hooks for quantization but don't currently support it. This is a valuable add, especially if the hardware isn't capable of loading the whole model in memory.

Ask

Integrate model quantization for all models.