Evidence organised for decisions, not volume.

Observed

The approved runs expose measurable resource behavior with source rows and bounded interpretation.

Supports

Public aggregate memory and smoke-loss guardrail curves across the approved experimental scales.

Does not support

These curves do not establish dense-baseline superiority or production performance.

Next decisive evidence

Repeat under a frozen current stack with matched controls before any broader comparison or performance use.

Evidence revision 6 July 2026

Decision relevance: research · diligence

Observed

Across 20 measurements per arm, direct block-region counting retained the tested plan identity and reduced mean 65,536-token plan-construction time from 784.49 ms to 7.99 ms on the disclosed host.

Supports

Exact plan-identity checks and host-local construction timing for one sparse-attention planning component.

Does not support

This is component-level engineering evidence. It does not establish end-to-end speed, model quality, memory or energy savings, cross-host performance, or superiority.

Next decisive evidence

Integrate the component into a matched quality-bearing end-to-end evaluation before expanding its performance meaning.

Evidence revision 16 July 2026

Decision relevance: research · engineering

Observed

Tiny and Small completed the declared ladder through 524,288 tokens, while Medium completed every safely admitted point through 131,072; 4,991 metric rows closed with zero observatory errors.

Supports

A replicated current-host context ladder across three CPU-only synthetic candidates, with cache-representation counts, semantic diagnostics, and admission boundaries.

Does not support

The history is synthetic all-zero K/V. These results do not establish language quality, generalisation, energy, whole-process memory, cross-host performance, end-to-end throughput, or superiority.

Next decisive evidence

Clear final assembly CI, then advance from synthetic all-zero K/V to representative histories and quality-bearing workloads.

Evidence revision 16 July 2026

Decision relevance: research · investment

Observed

All 28 donor layers produced balanced expert assignments, K32 matched five frozen engineering cases exactly, and all 112 protected trace units completed across four isolated roles.

Supports

Real Qwen3 0.6B fracture, exact K32 engineering reconstruction, and a completed protected four-way trace for material-sparsity alignment.

Does not support

The bounded alignment path is executable, but no complete aligned checkpoint is promoted; K8/K16/K24 quality, savings, broader generalisation, and production readiness remain unevaluated.

Next decisive evidence

Use the retained negative geometry to test refracture and recovery, then rerun the material-sparsity ladder under quality-bearing controls.

Evidence revision 12 July 2026

Decision relevance: research · investment · diligence

Observed

Small controlled probes provide measurement signals suitable for engineering decisions and follow-up experiments.

Supports

Step counts, throughput ranges, parameter footprint, and pair-work reduction from bounded probes.

Does not support

Probe-scale behavior is not a substitute for matched end-to-end production benchmarks.

Next decisive evidence

Regenerate from the repaired exact artifact identity and close or retain the phase-41 contract-discovery failure.

Evidence revision 6 July 2026

Decision relevance: research · engineering

Observed

The training continuation path is instrumented and produces governed stability evidence.

Supports

Loss signals, stability controls, generation gates, reload behavior, and token-scale receipts.

Does not support

Generation quality and broad model capability remain under review and are not mature public claims.

Next decisive evidence

Run and seal a current exact-resume continuation with frozen evaluation and generation review.

Evidence revision 6 July 2026

Decision relevance: research

Observed

The integrated training path runs with observable safety and quality signals.

Supports

Bounded throughput, loss, safety-counter, warning, and claim-boundary evidence from the full training path.

Does not support

Current evidence does not approve commercial product readiness or unrestricted generation claims.

Next decisive evidence

Select one exact run and bind its metrics, receipts, model, data, configuration, and review before refresh.

Evidence revision 6 July 2026

Decision relevance: research · diligence

Observed

The scale matrix exposes a concrete engineering checkpoint with explicit dimensions and receipts.

Supports

Active parameters, evaluation loss, memory peak, and throughput-ratio results at the approved scale point.

Does not support

A single bounded scale matrix does not prove performance superiority across hardware or workloads.

Next decisive evidence

Repeat the matrix under one frozen current profile and matched control before current-stack use.

Evidence revision 6 July 2026

Decision relevance: research · engineering

Observed

At the recorded boundary, the engineering inventory and exact-reconstruction anchor were visible while sparse alignment remained the active proof gate.

Supports

A timestamped repository-scale, evidence-inventory, and nine-claim presentation snapshot retained for historical traceability.

Does not support

This snapshot is not the live claim register. Inventory scale and completed engineering gates do not establish material-sparse quality, performance, or production readiness.

Next decisive evidence

Regenerate state of play after the claim, phase, fracture, and evidence releases close at one approved exact SHA.

Evidence revision 12 July 2026

Decision relevance: diligence · research

Observed

Review controls are visible and machine-auditable, including unresolved and held items.

Supports

Proof inventory, public claim posture, release checklist, and risk-register status.

Does not support

A review-ready evidence pack is not equivalent to approval for public production inference.

Next decisive evidence

Rebuild the review packet and release checklist from the next approved evidence release.

Evidence revision 6 July 2026

Decision relevance: diligence · governance

Observed

Critical system surfaces are governed by explicit pass/fail controls rather than narrative assurance alone.

Supports

Configuration validation, critical gate coverage, phase-control checks, and API observability posture.

Does not support

Control coverage describes engineering discipline; it does not independently prove model capability.

Next decisive evidence

Rerun and close—or retain—the three failing raw-discovery validations, then publish an exact-SHA quality-gate pack with executed assembly CI.

Evidence revision 6 July 2026

Decision relevance: governance · diligence

Observed

The package records 38/41 raw phase discovery, 44 reconciled claim rows, a guarded 2,048K synthetic CPU candidate, and two component-geometry failures that narrow the next refracture and recovery experiments.

Supports

An exact-source, sanitized public bundle joining current phase and claim reconciliation with a 2M synthetic context candidate and two retained sparse-geometry negatives.

Does not support

Source-family CI passed, but final assembly CI did not start because of an external billing limit. The package does not promote language quality, sparse serving, end-to-end performance, production readiness, or commercial superiority.

Next decisive evidence

Restore CI availability and execute the exact final package commit, while retaining the three raw phase-discovery failures and all negative scientific results unless they genuinely close.

Evidence revision 28 July 2026

Decision relevance: research · investment · diligence · governance

Observed

The outlook turns architecture assumptions into inspectable resource scenarios for planning.

Supports

Projected active share, stored parameters, decode bandwidth, optimizer state, and weight-stream requirements.

Does not support

All values are projections, not measured deployment results, and depend on the published assumptions.

Next decisive evidence

Refresh assumptions from the approved current architecture and keep projections separate from measured results.

Evidence revision 6 July 2026

Decision relevance: investment · planning

Observed

Diligence artifacts and boundary controls are explicitly inventoried rather than implied.

Supports

Artifact coverage, claim posture, clean-room controls, security review, and controlled gates.

Does not support

Readiness for diligence is not a declaration that all product, legal, or release risks are closed.

Next decisive evidence

Rebuild diligence coverage from the next clean approved evidence release and retain every unresolved gate.

Evidence revision 6 July 2026

Decision relevance: diligence · investment · governance