Skip to main content
← Lab

countcontrol

confirmedPython · CPUModel behavior & reasoningJuly 6, 2026

the question

Can a small model output exactly the number of items you ask for?

what came out

Not without a counter -- it overshoots 'exactly 20' by tens and duplicates up to 44%. Numbering the list fixes it almost completely (McNemar p~1e-20). The counting analog of a running sum. Exact oracle.

method & receipts

  • Result: confirmed
  • Reproducible one script re-runs the whole thing from scratch.
  • Independently reviewed an adversarial review pass, kept in the repo.
  • Machine-readable claims every reported number, in a checkable file.
  • Tested — a correctness/benchmark suite ships alongside the code.

→ read the code and re-run it

github.com/v-code01/countcontrol