Projects · Assurance architecture · Research prototype · MEASURED

nanogpt-provenance

What this establishes

A tamper-evident provenance record with a verifier that re-derives a training run's result bit-for-bit from recorded inputs and reports VERIFIED, TAMPERED or UNKNOWN, never silently upgrading UNKNOWN.

What this does not establish

Cross-hardware re-derivation, which is untested and reported UNKNOWN. It does not establish code correctness, buggy-but-faithful code still verifies, nor result quality nor accountability.

Trusted base

SHA-256 collision-resistance; the verifier's own CPython, NumPy, operating system and CPU; and that the verifier's numeric environment reproduces the record's, which is the weakest link.

Method

The verifier re-runs training from the recorded configuration and compares the trajectory and final loss bit-for-bit under a matching numeric environment; a 12-case negative-control suite triggers each verdict state with a real tamper.

Result

On one machine, VERIFIED with final loss 0.10635284871660042 and fingerprint c6da865d, 12 of 12 controls; a re-derivation mismatch under a foreign environment returns UNKNOWN, not TAMPERED.

Reproduce

./run_all.sh

Toolchain

Python 3.14.4, NumPy 2.4.4 on Apple Accelerate BLAS

Verified at

repowazdogz-droid/nanogpt-provenance@732da8b1 (2026-07-20)

This pass. Re-verified this pass: run_all.sh was re-run from a fresh public clone at 732da8b1, reproducing the final loss and fingerprint bit-for-bit with 12 of 12 controls passing.

Independent reproduction

None known.

Prior art and positioning

Provenance and reproducible builds; tamper-evident logs. The contribution is re-derivation rather than integrity-recheck, with an honest UNKNOWN exactly where reproducibility ends.

Links

• Source: repowazdogz-droid/nanogpt-provenance
• Finding: A training run's final loss and trajectory re-derive bit-for-bit from its recorded inputs on one machine; cross-hardware re-derivation is left as UNKNOWN, and not claimed.