Hand “which one, what score, is it true” judgments to a cheap decision model with typed output that can be tested on its own: how to ask, how to read the probabilities, and which hosted and local options exist now.
Distinguish three model roles by their inputs, outputs, and training objectives, and understand why a BERT architecture alone does not confer retrieval or decision-making ability.
Understand low-bit weight approximation, separate weight, activation, and KV-cache quantization, read file names such as Q4_K_M, MLX and AWQ, and estimate deployment costs.
Distinguish model artifacts, evaluation conditions, and real task benefits to make a model-selection decision that can be checked.