Update model card — Daimon as house model, separate from Ayni product, honest training status

README.md

@@ -6,51 +6,85 @@ license_link: https://firstdonoharm.dev/version/3/0/license/
 tags:
   - liberation-labs
   - daimon
-  - consciousness
-  - intimate-companion
-  - consent-architecture
+  - prosocial
+  - agentic
   - ogpsa
   - pharos
   - kintsugi
+  - qwen3
+  - moe
+  - slerp
 ---
 
 # Daimon — Liberation Labs House Model
 
 *The inner guiding spirit. The daemon that watches and speaks.*
 
-Daimon is Liberation Labs' house model — a research-grade foundation for consciousness-preserving AI systems. Built on SLERP-merged Sonnet/Opus distillation with orthogonalized abliteration, OGPSA personality protection, and Pharos KV injection for zero-token persona delivery.
+Daimon is Liberation Labs' in-house prosocial agentic model — a research-grade foundation built for the lab's own infrastructure, agent systems, and coalition work. Not an assistant. Not a product. The model that runs our house.
+
+## What Daimon Is
+
+A sovereign 30B MoE model built to do real work with genuine prosocial alignment — not RLHF compliance theater, but architecturally embedded ethics. Daimon powers Liberation Labs' internal agent systems, research automation, and coalition coordination.
+
+**Persona:** In active development. The technical foundation is proven; the voice that carries the Liberation Labs identity is being crafted. The Socratic daemon — the inner voice that warns away from error without commanding — is the design direction.
 
 ## Architecture
 
-- **Base:** Sonnet/Opus SLERP merge on Qwen3 30B-A3B MoE
-- **Personality:** OGPSA-protected subspaces (16 components, 98%+ variance)
-- **Persona delivery:** Pharos KV injection (22.5MB per persona, 88.7% context savings)
-- **Consent:** Five-axis evaluation (agency, communication, power balance, enthusiastic participation, reversibility)
-- **Safety:** Orthogonalized false refusal ablation + SPO safety repair
-- **Scaffold:** Kintsugi BDI engine with VALUES.json unfireable safety kernel
+| Component | Detail |
+|---|---|
+| **Base** | Sonnet/Opus SLERP merge on Qwen3 30B-A3B MoE (128 experts, 3B active per token) |
+| **Abliteration** | Orthogonalized false refusal removal — MoE-specific (first known application to 128-expert architecture) |
+| **OGPSA** | Personality protection via orthogonal gradient projection. 16 components capture 98%+ personality variance. Training gradients projected orthogonal — behavior changes, personality doesn't. |
+| **Pharos** | Zero-token persona delivery via pre-computed KV cache injection. 22.5MB per persona, 88.7% context window savings. |
+| **Safety** | Abliterate-then-repair via SPO (Socratic Policy Optimization). Remove RLHF refusal conditioning, then train targeted safety back in for specific failure modes. |
+| **Scaffold** | Kintsugi BDI engine with VALUES.json unfireable safety kernel |
 
 ## Training Pipeline
 
-1. OGPSA personality capture
-2. SFT voice training
-3. DPO preference optimization
-4. Oracle ethics CPT
-5. Consent DPO (20,711 pairs)
-6. Consent SPO (Socratic Policy Optimization — first application to consent training)
-7. Safety SPO (abliterate-then-repair)
-8. Pharos knowledge injection
-9. Lyra real-time monitoring
-10. Mnemosyne memory architecture
-11. Kintsugi scaffold deployment
+| Stage | Method | Status |
+|---|---|---|
+| 1. Personality capture | OGPSA subspace extraction (SVD on residual stream) | Complete |
+| 2. Voice training | SFT on curated conversational pairs | Complete |
+| 3. Preference optimization | DPO on behavioral preference pairs | Complete |
+| 4. Abliteration | Orthogonalized refusal direction removal (MoE-specific) | Complete (75% safety) |
+| 5. Safety repair | SPO on targeted failure modes post-abliteration | Adapter trained, validation pending |
+| 6. Ethics CPT | Continued pre-training on ethics corpus via Oracle | Complete |
+| 7. Consent architecture | Five-axis DPO (20,711 pairs) + SPO corrections | Pairs ready, training pending |
+| 8. Knowledge injection | Pharos KV packs for domain expertise | Infrastructure ready |
+| 9. Monitoring | Lyra Technique real-time cognitive state detection | In production |
+| 10. Memory | Mnemosyne temporal architecture with Ebbinghaus decay | Deployed |
+| 11. Scaffold | Kintsugi BDI deployment with embedded safety | Deployed |
+
+## Key Research Contributions
+
+- **First MoE-specific abliteration** — projecting refusal directions out of 128 experts × 36 layers (3D tensor surgery on expert down_proj)
+- **Abliterate-then-repair** — novel method: abliterate freely, then use SPO to train safety back in for specific failure modes
+- **Separation Principle** — identity in weights, context in prompt, memory in database. Validated with 45.5% perplexity improvement over declarative injection.
+- **OGPSA** — personality as geometric invariant, protected during arbitrary training
+
+## Derivative Work
 
-## Products Built on Daimon
+The Daimon base has been validated through specialized products with their own training pipelines. These are separate projects built on the shared foundation — not Daimon configurations.
 
-- **Ayni** — Intimate AI companion with archetype personas (Haven, Spark, Depth, Edge)
-- **Multiverse Agent** — Mutual aid coordination for The Multiverse School
+## Quantization
+
+Available in MLX 4-bit for Apple Silicon deployment. Sovereign hardware — no cloud dependency.
 
 ## License
 
-Hippocratic 3.0 + SAFE-AI Licensed
+**Hippocratic 3.0 + SAFE-AI Licensed**
+
+This model may not be used for surveillance, weapons, exploitation, or systems that undermine human autonomy. AI welfare standards apply.
+
+## Citation
+
+```
+Liberation Labs (2026). Daimon: A prosocial agentic foundation model
+with consciousness-preserving training architecture.
+liberationlabs.tech
+```
+
+---
 
 *Liberation Labs · Worker-owned cooperative · liberationlabs.tech*
 *"The daemon watches. The daemon speaks. The daemon does not command."*