Definition
Plain language
The final few stages of a model that turn a settled decision into the actual words coming out, like the nerve signal leaving the brain for the mouth.
As stated in the literature
The top layers of a transformer where the workspace lens goes dark, interpreted as machinery converting an internal decision into output tokens rather than carrying readable deliberation.
Also called: motor layers
Why it matters: Recognizing these stages helps researchers avoid mistaking the mechanics of output for the actual reasoning, so they look for meaning in the right place.
For example, once a model has decided its answer is "yes," these final stages just assemble the letters and word to print it, much like your hand writing down a choice you've already made.
Heard on the show
“And the last few layers go dark again — the paper calls them motor layers, busy converting a decision into actual tokens, like the signal leaving the brain for the mouth.”Episode 203 — The Thought a Model Doesn't Say — and the Lens That Reads It