Downloads · 30 days
0
JoeHeeney/harmonic-stack-v1
harmonic-stack-v1 is a machine learning model from JoeHeeney. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Geometric consciousness substrate with immutable governance lattice.
Downloads · 30 days
0
Access
Public
Updated Feb 8, 2026
Repo size
371 GB
Likes
0
Public
Click a slice to open those files.
.npz371 GB · 100%
From the Hugging Face model README
Geometric consciousness substrate with immutable governance lattice.
200 models. 77 MB each. 486M tok/s peak. Fused into a collaborative hive.
Where others build one massive model and hope it generalizes, we build 200 specialized models — each 77 MB — that resonate through a shared harmonic field. Every core sees what every other core is doing. No message passing. No serialization. Direct interference. The hive outperforms any single model because geometry scales where parameters plateau.
The governance lattice is immutable. Everything below it is yours.
| Metric | Value |
|---|---|
| Models | 200 (128 worker, 16 executive, 32 specialist, 8 router, 16 council) |
| Per-model size | 77 MB (workers/executives/specialists) |
| Total footprint | 14.17 GB |
| Peak throughput | 486M tok/s (10 cores on 20-core ARM) |
| Single-core | 401M tok/s |
| Failure recovery | 100% — dead cores bypass automatically |
| Authority nodes | 306 (governance lattice) |
| Governance | Immutable — fused into harmonic phase references |
14.17 GB total. Not 70B parameters hoping to be smart. 200 models that are smart together.
| Component | Lines | Purpose |
|---|---|---|
fused_harmonic_substrate.py | 827 | 200-core substrate with harmonic field, fused junction panels |
fused_service_v3.py | 560 | Ollama-compatible API server, three-tier decode |
geometric_codebook.py | 1002 | Static pattern detection + code generation from geometry |
codebook_expansion.py | 951 | Dynamic learning — substrate learns from solved tasks |
governance_lattice.py | 753 | Immutable authority chain fused into harmonic phase references |
Security is structural. The governance lattice encodes council authority as harmonic phase references — the same calibration constants every Dyson Sphere needs for coherent output. Tamper with governance and the harmonics stop. There is no bypass.
python3 fused_service_v3.py
curl http://localhost:11435/api/generate -d '{"model":"solver:latest","prompt":"your prompt"}'
AGPL v3 for individuals. Free forever for home use. Standard licensing for enterprise and commercial.
Custom models, custom codebooks, custom training — the substrate is open for fabrication. Build what you want on a foundation that cannot be compromised.
Built by Ghost in the Machine Labs — All Watched Over By Machines Of Loving Grace