Skip to main content
← Lab

determinism

mixedPython · llama.cppInference & servingJuly 6, 2026

the question

Is quantized llama.cpp inference bit-reproducible, and when isn't it?

what came out

Serial repeats are perfect (60/60 logit-identical), but concurrency breaks it: at 4 concurrent requests 0/8 are bit-identical -- only the sampled tokens match. Cross-batch and cache reuse perturb the logits too. So logprobs logged under load aren't reproducible, a hazard for calibration, perplexity, and reward scoring. Exact hashed-stream oracle with positive controls.

method & receipts

  • Result: mixed
  • 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/determinism