[ethbr0]

My takeaway is that the more we turn to larger and larger models (developed by third parties) that obscure their underlying nuances, the more risky swapping models in a particular domain becomes.

Cue MRE/MLOps job listings.