Doesn't seem relevant here. TurboQuant isn't a domain-specific technique like the BL is talking about, it's a general optimisation for transformers that helps leverage computation more effectively.