A city flax-retting pond office needs three text jobs: a stack that only reads a note, a stack that only writes a new note, and a stack that reads one note then writes another. Which transformer layouts match those three roles?
Select an answer to reveal the explanation.
Short Explanation
Picture three desks at city hall. One only reads memos (encoder-only), one only drafts them (decoder-only), and one reads a memo then writes a reply (encoder–decoder). Match the job to the layout.
Full Explanation
Encoder-only stacks specialize in understanding input text. Decoder-only stacks specialize in generating text. Encoder–decoder stacks encode one sequence and decode another, which fits read-then-write. Collapsing those roles into one layout, or swapping encoder and decoder roles, misses how transformers are used as LLM building blocks.