Fail-closed baseline pairing with lower-memory retries

Release: InvarLock 0.3.4 - Required pairing evidence and chunked adapter snapshots

October 19, 2026 revision: Corrected the historical helper description to bootstrap, scheduler/queue utilities, and model creation tooling; retained the release’s certificate terminology.

Highlights

• CI/Release baseline pairing now fails closed with strict parity checks.
• Certificate generation requires paired-windows evidence (non-perfect pairing is rejected).
• Chunked snapshot/restore and B200 validation workflow helpers reduce retry and long-run friction.

This release makes baseline pairing a checked contract instead of a convention. If a CI/release run says it is paired to a baseline, it now has to carry valid evaluation_windows evidence, and the system enforces dataset, tokenizer, and masking parity instead of letting ambiguous comparisons slip through.

Certificate generation is equally strict: paired_windows evidence is required, and non-perfect pairing is rejected. That gives reviewers a clearer answer to the most important question: when a certificate says it is paired, it really is paired.

On the practical side, chunked HF adapter snapshot/restore helps reduce peak memory during retries, and the new B200 validation workflow helpers (bootstrap, scheduler/queue utilities, and model creation tooling) make longer evidence runs easier to operate without weakening the pairing rules.

Sources

Website documentation explains the maintained workflow. The changes described here belong to InvarLock v0.3.4; the tagged release record preserves that historical scope.

• Documentation
• Tagged v0.3.4 changelog