https://invarlock.aiRelease notes, tutorials, and updates from InvarLock.https://invarlock.ai/blog/release-0-11-0https://invarlock.ai/blog/release-0-11-0Tue, 16 Jun 2026 00:00:00 GMTInvarLock 0.11.0 makes human reports more consistent, separates guard warnings from hard failures, and expands the public evidence surface across model families.https://invarlock.ai/blog/evidence-packs-not-screenshotshttps://invarlock.ai/blog/evidence-packs-not-screenshotsMon, 15 Jun 2026 00:00:00 GMTA screenshot can communicate a result. An evidence pack can be inspected, checked, and re-verified later. That is the difference between presentation and portable evidence.https://invarlock.ai/blog/runtime-manifests-and-why-provenance-must-travel-with-the-resulthttps://invarlock.ai/blog/runtime-manifests-and-why-provenance-must-travel-with-the-resultMon, 08 Jun 2026 00:00:00 GMTA strong evaluation result should carry its runtime provenance with it. In InvarLock, that means the runtime manifest travels next to the report and is rechecked by invarlock verify.https://invarlock.ai/blog/release-0-10-0https://invarlock.ai/blog/release-0-10-0Wed, 03 Jun 2026 00:00:00 GMTInvarLock 0.10.0 makes public evidence packs more explicit, adds signer-authenticity checks, and expands optional quantized-subject adapter validation.https://invarlock.ai/blog/what-belongs-in-evaluation-report-jsonhttps://invarlock.ai/blog/what-belongs-in-evaluation-report-jsonMon, 01 Jun 2026 00:00:00 GMTAn evaluation report is strongest when it is treated as a stable evidence contract: a small required core, meaningful optional blocks, and a clear boundary around what still lives outside the JSON.https://invarlock.ai/blog/calibration-is-the-product-surface-not-a-side-utilityhttps://invarlock.ai/blog/calibration-is-the-product-surface-not-a-side-utilityMon, 25 May 2026 00:00:00 GMTCalibration is not just analysis around the product. It changes how thresholds are derived, when correction paths may turn on, and which policy values later govern reports.https://invarlock.ai/blog/release-0-9-0https://invarlock.ai/blog/release-0-9-0Mon, 25 May 2026 00:00:00 GMTInvarLock 0.9.0 adds strict assurance mode, fail-closed verifier checks, runtime provenance guidance, and maintainer evidence gates for release review.https://invarlock.ai/blog/from-sweep-outputs-to-tier-policyhttps://invarlock.ai/blog/from-sweep-outputs-to-tier-policyMon, 18 May 2026 00:00:00 GMTCalibration becomes operational when sweep artifacts end in reviewable YAML patches that later appear as resolved runtime policy in reports.https://invarlock.ai/blog/variance-enablement-should-be-evidence-gatedhttps://invarlock.ai/blog/variance-enablement-should-be-evidence-gatedMon, 11 May 2026 00:00:00 GMTVariance equalization is stronger when it must earn enablement through predictive evidence, explicit tier knobs, and report-visible provenance.https://invarlock.ai/blog/null-sweeps-as-threshold-derivation-not-tuning-folklorehttps://invarlock.ai/blog/null-sweeps-as-threshold-derivation-not-tuning-folkloreMon, 04 May 2026 00:00:00 GMTThresholds are stronger when they come from measured null behavior and end in a policy patch, not from knob-tuning folklore.https://invarlock.ai/blog/the-minimum-evidence-surface-for-trustworthy-weight-edit-resultshttps://invarlock.ai/blog/the-minimum-evidence-surface-for-trustworthy-weight-edit-resultsMon, 27 Apr 2026 00:00:00 GMTA trustworthy weight-edit result needs more than a benchmark delta. It needs a bounded claim, an exactly paired comparison, and verification that rejects incomplete evidence.https://invarlock.ai/blog/release-0-8-0https://invarlock.ai/blog/release-0-8-0Thu, 23 Apr 2026 00:00:00 GMTInvarLock 0.8.0 moves the public bundle surface to evidence packs, pins docs to versioned release paths, and makes container-vs-host runtime provenance explicit across evaluate and verify.https://invarlock.ai/blog/fail-closed-verification-for-weight-edit-evaluationhttps://invarlock.ai/blog/fail-closed-verification-for-weight-edit-evaluationMon, 20 Apr 2026 00:00:00 GMTA verifier is only useful if it rejects incomplete evidence. InvarLock's verification path is designed to stop stronger claims when the evidence bundle is missing or inconsistent.https://invarlock.ai/blog/release-0-7-2https://invarlock.ai/blog/release-0-7-2Wed, 15 Apr 2026 00:00:00 GMTInvarLock 0.7.2 simplifies the public release surface around immutable source tags plus the PyPI wheel and sdist, with docs and verification gates aligned around that path.https://invarlock.ai/blog/why-paired-evaluation-beats-before-after-benchmarkshttps://invarlock.ai/blog/why-paired-evaluation-beats-before-after-benchmarksMon, 13 Apr 2026 00:00:00 GMTA model-edit benchmark number is only as strong as the comparison behind it. Pairing makes the comparison inspectable.https://invarlock.ai/blog/release-0-7-1https://invarlock.ai/blog/release-0-7-1Sun, 12 Apr 2026 00:00:00 GMTInvarLock 0.7.1 makes wheel-only verify/report workflows first-class, ships a public contract bundle, and tightens supply-chain and release-validation gates.https://invarlock.ai/blog/release-0-7-0https://invarlock.ai/blog/release-0-7-0Thu, 09 Apr 2026 00:00:00 GMTInvarLock 0.7.0 adds first-class GPT-OSS support, pilot Ministral 3 8B/14B presets, and a CUDA-capable attested runtime path for GPU hosts.https://invarlock.ai/blog/what-invarlock-actually-claimshttps://invarlock.ai/blog/what-invarlock-actually-claimsMon, 06 Apr 2026 00:00:00 GMTA narrow claim can be stronger than a broad one. InvarLock is about auditable regression risk from weight edits, not general model safety.https://invarlock.ai/blog/release-0-6-0https://invarlock.ai/blog/release-0-6-0Sat, 04 Apr 2026 00:00:00 GMTInvarLock 0.6.0 adds a shipped Gemma 4 E2B text lane, phase-1 multimodal evaluation, and a unified `--assurance attested|trusted-local` workflow.https://invarlock.ai/blog/release-0-5-1https://invarlock.ai/blog/release-0-5-1Thu, 02 Apr 2026 00:00:00 GMTInvarLock 0.5.1 adds a push-gated tiny attested smoke lane, a scheduled GPT-2 canary lane, and package-native Ed25519 evidence pack signatures.https://invarlock.ai/blog/release-0-5-0https://invarlock.ai/blog/release-0-5-0Wed, 25 Mar 2026 00:00:00 GMTInvarLock 0.5.0 adds offline release-verification bundles, package-native evidence-pack verification, and a simplified public CLI centered on evaluate, verify, and report.https://invarlock.ai/blog/release-0-4-0https://invarlock.ai/blog/release-0-4-0Sat, 14 Mar 2026 00:00:00 GMTInvarLock 0.4.0 stabilizes contracts around policies, evidence packs, and evaluation provenance while tightening verification, CI, and coverage enforcement.https://invarlock.ai/blog/release-0-3-12https://invarlock.ai/blog/release-0-3-12Fri, 27 Feb 2026 00:00:00 GMTSplit-module coverage thresholds now protect critical CLI/reporting paths while config, plugin, report, overhead, and observability edge cases fail closed more reliably.https://invarlock.ai/blog/release-0-3-11https://invarlock.ai/blog/release-0-3-11Thu, 12 Feb 2026 00:00:00 GMTA focused hardening release: safer AWQ plugin discovery, stronger quantization clipping behavior, and broader report-schema acceptance for edge payloads.https://invarlock.ai/blog/release-0-3-10https://invarlock.ai/blog/release-0-3-10Sun, 08 Feb 2026 00:00:00 GMTEvidence packs add new showcase and evidence artifacts, while CI and release flows become more deterministic and easier to validate repeatedly.https://invarlock.ai/blog/release-0-3-9https://invarlock.ai/blog/release-0-3-9Tue, 03 Feb 2026 00:00:00 GMTA stability-focused release: cleaner report output, safer offline evidence-pack flows, and CI/test hardening after the report rename.https://invarlock.ai/blog/release-0-3-8https://invarlock.ai/blog/release-0-3-8Mon, 02 Feb 2026 00:00:00 GMTA terminology reset (report/evaluate), stricter evidence-pack verification, and a clean upgrade path for Hugging Face Transformers v5.https://invarlock.ai/blog/release-0-3-7https://invarlock.ai/blog/release-0-3-7Thu, 22 Jan 2026 00:00:00 GMTAdapters move to role-based routing, evidence packs become easier to inspect (v2 layout), and reporting output gets a readability upgrade.https://invarlock.ai/blog/release-0-3-6https://invarlock.ai/blog/release-0-3-6Tue, 13 Jan 2026 00:00:00 GMTReports now record and enforce estimator measurement contracts under CI/release profiles, and evidence pack suites can cleanly split calibration vs execution.https://invarlock.ai/blog/release-0-3-5https://invarlock.ai/blog/release-0-3-5Fri, 02 Jan 2026 00:00:00 GMTEvidence packs gain a deterministic bash test suite and better runtime helpers, window selection becomes stable/offline, and perplexity runs get safer around bad token IDs.https://invarlock.ai/blog/release-0-3-4https://invarlock.ai/blog/release-0-3-4Sun, 28 Dec 2025 00:00:00 GMTCI/release baseline pairing is fail-closed (pairing evidence is required), and adapters reduce peak memory during retries via chunked snapshot/restore.https://invarlock.ai/blog/release-0-3-3https://invarlock.ai/blog/release-0-3-3Sun, 21 Dec 2025 00:00:00 GMTToken-weighted paired bootstrap lands across the pipeline, strictness toggles expand, and CI/release pairing expectations become explicit and enforceable.https://invarlock.ai/blog/release-0-3-2https://invarlock.ai/blog/release-0-3-2Sun, 14 Dec 2025 00:00:00 GMT`invarlock calibrate` arrives, determinism utilities mature, and regression harness + golden tracking help prevent silent policy drift.https://invarlock.ai/blog/release-0-3-1https://invarlock.ai/blog/release-0-3-1Wed, 10 Dec 2025 00:00:00 GMTFixes a GPU memory leak during reload fallback, hardens B200 scripts, and adds practical controls for acceptance ranges and overhead measurement.https://invarlock.ai/blog/release-0-3-0https://invarlock.ai/blog/release-0-3-0Fri, 05 Dec 2025 00:00:00 GMTFirst-class quantization metadata, safer device movement across quantized models, auto-routing based on checkpoint info, and major test coverage expansion.https://invarlock.ai/blog/release-0-2-0https://invarlock.ai/blog/release-0-2-0Mon, 01 Dec 2025 00:00:00 GMTThe initial public release on GitHub and PyPI: core evaluate pipeline, guard chain, schema v1, and the first docs/CLI surface.https://invarlock.ai/blog/welcome-to-invarlockhttps://invarlock.ai/blog/welcome-to-invarlockSun, 30 Nov 2025 00:00:00 GMTA quick introduction to InvarLock: evaluate LLM weight edits with statistical guarantees and auditable evidence packs.