Skip to main content
← Lab

samplercost

confirmedPython · llama.cppInference & servingJuly 6, 2026

the question

Which sampling method actually costs latency, and does sampler order matter?

what came out

top-p is the costly one; top-k and min-p are near free; putting top-k before top-p removes top-p's cost (it shrinks the candidate set first); and top-p's overhead is larger for the smaller model. All four pre-registered predictions held.

method & receipts

  • Result: confirmed
  • Pre-registered the prediction was written down and committed before the run.
  • 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/samplercost