Projection scenarios
Planning scenarios, not measured deployments.
Projections and diligence
Resource outlook
Projected active share, stored parameters, decode bandwidth, optimizer state, and weight-stream requirements.
Recorded finding: The outlook turns architecture assumptions into inspectable resource scenarios for planning.
On this page
Currency and next evidence gate
Decision signals
Key metrics and controls
The recorded values, with their controls and scope.
Lowest projected active share
Derived from stated architecture assumptions.
Largest projected bandwidth ratio
Projected active versus same-size dense weight stream.
Largest stored-parameter scenario
Projection envelope only.
Source-linked visual evidence
What the evidence shows
Open any chart’s data table to inspect the exact values, or download the CSV.
Projection
Projected weight stream
Projected active weight stream compared with the same-size dense stream.
Finding: The projection estimates materially smaller active streams under its published assumptions.
Projection A
Projection B
View source values
Scroll sideways if any columns are out of view.
| Group | Projected active | Same-size dense |
|---|---|---|
| Projection A | 150.6 GiB | 1,500.6 GiB |
| Projection B | 420.6 GiB | 5,820.6 GiB |
Method
How to read this result
- Values are calculated from the public resource-audit projection envelope.
- The scenarios do not measure model quality, training memory, serving latency, or token throughput.
What this result establishes
All values are projections, not measured deployment results, and depend on the published assumptions.
- Projection math only; measured scale-ladder evidence and resource-audit projections must remain separate.
- Anonymous projection envelopes only; raw architecture inputs, scenario names, source paths, command names, script names, and mechanism knobs are not published.
- Parameter counts are normalized to M.
- Active-weight stream estimates are single-user decode projection signals, not throughput, training-memory, model-quality, production-readiness, or external reproduction claims.
- Global training, scoped residency, and optimizer-state estimates remain separate claims.
Inspection and reuse
Downloads and provenance
Source data
Machine-readable summary
The recorded JSON artifact is staged from the governed public evidence pack during the build. Its currency notice determines whether it may be read as current.
Open source JSONProvenance
Build and evidence receipt
Use the public build receipt to compare the deployed site source, evidence digest, claim posture, and inference release state.
Open build receiptContinue reviewing
Related evidence
measuredpartial
Historical- 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.
Decision relevance: research · engineering
Review evidencecontrolreviewed
Refresh required- 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.
Decision relevance: diligence · investment · governance
Review evidence