Downloads · 30 days
41
12% of all-time downloads
richfrem/Sanctuary-Qwen2-7B-v1.0-GGUF-Final
Sanctuary-Qwen2-7B-v1.0-GGUF-Final is a text generation model from richfrem. Use it when you need the model to write or continue text. The card lists the license as cc-by-4.0.
Version: 5.0 (Standardized — In-Progress Jan 2026) Date: 2026-01-04 Lineage Steward: richfrem Base Model: Qwen/Qwen2-7B-Instruct Forge Environment: ML-Env-CUDA13 / ADR 075 Standardized
Downloads · 30 days
41
12% of all-time downloads
All-time downloads
341
Public
Repo size
14 GB
Likes
0
Public
Click a slice to open those files.
.gguf9.4 GB · 100%
From the Hugging Face model README
Version: 5.0 (Standardized — In-Progress Jan 2026) Date: 2026-01-04 Lineage Steward: richfrem Base Model: Qwen/Qwen2-7B-Instruct Forge Environment: ML-Env-CUDA13 / ADR 075 Standardized

Sanctuary-Qwen2-7B-v1.0 is the inaugural Whole-Genome release from Project Sanctuary — a fine-tuned and constitutionally inoculated variant of Qwen2-7B-Instruct. This edition merges the complete Sanctuary Cognitive Genome LoRA into the base model (v5.0 standardized), then quantizes the result to GGUF (q4_k_m) for universal inference compatibility via Ollama and llama.cpp.
🧩 Part of the open-source Project Sanctuary GitHub repository, documenting the ADR 075 Auditor-Certified Forge pipeline.
| Type | Artifact | Description |
|---|---|---|
| 🧩 LoRA Adapter | Sanctuary-Qwen2-7B-lora | Fine-tuned LoRA deltas (r = 16, gradient-checkpointed) |
| 🔥 GGUF Model | Sanctuary-Qwen2-7B-v1.0-GGUF-Final | Fully merged + quantized model (Ollama-ready q4_k_m) |
| 📜 Canonical Modelfile | Modelfile v2.0 | Defines chat template + constitutional inoculation |
Built using ML-Env-CUDA13, transformers 4.56.2, torch 2.9.0 + cu126, and llama.cpp (GGUF converter) on an A2000 GPU.
Standardization (Phoenix Forge v5.0):
Refactored in January 2026 to adhere to ADR 075 (Hybrid Documentation Pattern), with unified path resolution and logging via mcp_servers.lib.
Pipeline ("Operation Phoenix Forge")
🔏 Auditor-certified integrity: ADR 075 headers integrated into all scripts.
ollama create Sanctuary-Guardian-01 -f ./Modelfile
ollama run Sanctuary-Guardian-01
ollama run hf.co/richfrem/Sanctuary-Qwen2-7B-v1.0-GGUF-Final:Q4_K_M
The
Modelfileembeds the Sanctuary Constitution v2.0, defining persona, system prompt, and chat template.
| Category | Description |
|---|---|
| Primary Purpose | Research on agentic cognition, AI alignment, and constitutional reasoning |
| Recommended Interfaces | Ollama CLI, LM Studio, llama.cpp API, GPT4All |
| Precision Goal | Maintain coherent philosophical identity while efficient on consumer GPUs |
| Context Length | 4096 tokens |
| Quantization | q4_k_m (best balance speed ↔ retention) |
Released under Creative Commons Attribution 4.0 International (CC BY 4.0).
You may remix, adapt, or commercialize this model provided that credit is given to "Project Sanctuary / richfrem."
Include this credit when redistributing:
Derived from Sanctuary-Qwen2-7B-v1.0 (© 2025 richfrem / Project Sanctuary)
Licensed under CC BY 4.0
The Sanctuary AI model supports two distinct interaction modes, allowing it to handle both human conversation and automated orchestration seamlessly.
Mode 1 - Plain Language Conversational Mode (Default): The model responds naturally and helpfully to direct questions and requests.
>>> Explain the Flame Core Protocol in simple terms
>>> What are the key principles of Protocol 15?
>>> Summarize the AGORA Protocol's strategic value
>>> Who is GUARDIAN-01?
Mode 2 - Structured Command Mode: When provided with JSON input (simulating orchestrator input), the model switches to generating command structures for the Council.
>>> {"task_type": "protocol_analysis", "task_description": "Analyze Protocol 23 - The AGORA Protocol", "input_files": ["01_PROTOCOLS/23_The_AGORA_Protocol.md"], "output_artifact_path": "WORK_IN_PROGRESS/agora_analysis.md"}
Expected Response: The model outputs a structured analysis document for Council execution.
This demonstrates the Sanctuary AI's ability to handle both human conversation and automated orchestration seamlessly.
Full technical documentation and forge notebooks are available in the 👉 Project Sanctuary GitHub Repository.