Skip to content
Technology How it works Breeder — Hyperion Burner — Aegis Burner — MetroVolt AI-Native Architecture Magnets Fuel cycle Safety Roadmap
Solutions AI & Data Centers Defense & Government Grid & Baseload Neutron Detection Quantum
Learn Technical Library
Proof Publications Whitepapers Technical Library Open Science & Reproducibility The Honest Gates
Company About / Mission Leadership Environment Health & Safety Investors Careers Press Contact
3D Model
AI Architecture › MLOps & Learning
MLOps & Learning

Model Versioning and Lineage

Each model artifact carries a full provenance chain — data, code, seed, environment, and parent model — so any deployed controller can be reconstructed and audited exactly.

STRATEGY / SLOW ▲ ▼ MICROSECOND REAL-TIMEL7Ecosystem & Strategytelemetry ▲ control ▼open ▸L6Experience & Visualizationtelemetry ▲ control ▼open ▸L5Applications & Copilotstelemetry ▲ control ▼open ▸L4Orchestrationtelemetry ▲ control ▼open ▸L3Twin Modeling & AItelemetry ▲ control ▼open ▸L2Data Fabrictelemetry ▲ control ▼open ▸L1Control Planetelemetry ▲ control ▼open ▸L0Foundationtelemetry ▲ control ▼open ▸PHYSICAL S.M.A.R.T. GENERATOR PLANTBREEDER · HYPERION1R0 1.2 m · A 2.5 · 16.84 T · δ −0.30BURNER · TANDEM MIRROR2317 T throat · 26.49 T plug · fₙ 5.44% · DEC1 center stack + plasma · 2 high-field plug · 3 expander → direct converterCOLOR GRAMMAR strategy AI-workflow infra/data models reactor/DECLINE SEMANTICStelemetry (µs)controlKRONOS FUSION ENERGYAI-NATIVE S.M.A.R.T. GENERATORMASTER BLUEPRINTSHEET 01REV. 2026-08L0-L7 · 2 MACHINES
The AI-Native S.M.A.R.T. Generator Master Blueprint — eight layers (L0→L7), one control stack, wired to both machines. Telemetry rises in microseconds; control descends the same path.

A model is more than weights

A deployable Kronos model is a bundle: the weights, the exact training code revision, the pinned dataset hash, the random seeds, the container image digest, the hyperparameters, and a pointer to its parent model. Version the bundle, not just the weights. This is what lets a reviewer certify not a black box but a fully specified, rebuildable object.

Versioning follows semantic rules tied to behavior, not to file changes. A patch bump is a retrain on the same architecture and feature set; a minor bump changes features or adds data regimes; a major bump changes architecture, output contract, or safety envelope. Any major bump forces a full re-validation and a fresh shadow and canary cycle before it can act.

The lineage chain

python
model_card_meta = {
  'name': 'breeder/equilibrium-surrogate',
  'version': '4.1.0',
  'parent': '4.0.2',
  'dataset': 'sha256:9f2c...',
  'code': 'git:8ad91c', 'image': 'sha256:...', 'seed': 20291103,
  'validation': 'report:sha256:...',
  'state': 'STAGING'      # PROD only after gates + canary
}

Every artifact in the registry carries this chain, and the chain is append-only. When a model is rolled back, the record is not deleted; the superseding event is appended, so the operational history of which model held authority at which time on which machine is permanently reconstructable.

Content reviewed August 2026 · design-and-simulation stage