A municipal pigeon-loft log runs for many pages. Staff ask why a modern language model uses a transformer instead of stepping through the log with a long recurrent chain. What is the architectural reason?
Select an answer to reveal the explanation.
Short Explanation
Think of the loft log as a long hallway. A recurrent chain walks door to door; attention can glance across many doors at once. That whole-sequence view is why transformers became the LLM building block.
Full Explanation
Recurrent models process tokens sequentially and struggle to carry long-range context efficiently. Transformer self-attention lets each position weigh others across the sequence without that strict step-by-step walk. The advantage is architectural—parallel attention over the note—not a claim that recurrence never existed or that transformers abandon text.