---

language:

• en

license: apache-2.0

tags:

• cybersecurity
• scada
• iot
• esp32
• gguf
• qwen2.5-coder
• llama.cpp
• ollama

base_model: Qwen/Qwen2.5-Coder-32B-Instruct

pipeline_tag: text-generation

---

REGGIE-V6 (32B) — Sovereign Industrial Cybersecurity & Physical Telemetry Engine







REGGIE-V6-32B is an autonomous cybersecurity, industrial infrastructure, and hardware-sentinel LLM. Built on top of Qwen/Qwen2.5-Coder-32B-Instruct, it was fine-tuned with custom low-rank adapters (LoRA) and quantized to multiple GGUF formats using a native sovereign C++ streaming pipeline.

Read the full technical architectural journey and benchmark paper:

👉 **Read the Full Technical White Paper & Journey

**

👉 Raw White Paper Markdown

👉 GitHub Repository

---

⚡ Quantization Matrix & Hardware Recommendations

| File | Size | BPW | Recommended Hardware | Ollama 1-Click Command |

| :--- | :--- | :--- | :--- | :--- |

| reggie-v6-32b-Q3_K_M.gguf | 15.9 GB | 3.89 | 16 GB VRAM GPUs (RTX 4060 Ti 16GB, RTX 4080, 16GB Mac) | ollama run hf.co/Bigchilli1978/reggie-v6-32b-GGUF:reggie-v6-32b-Q3_K_M.gguf |

| reggie-v6-32b-Q4_K_M.gguf | 19.9 GB | 4.85 | 24 GB VRAM GPUs (RTX 3090 / 4090, 32GB+ Mac) | ollama run hf.co/Bigchilli1978/reggie-v6-32b-GGUF:reggie-v6-32b-Q4_K_M.gguf |

---

🚀 Quickstart with Ollama

Run directly on local hardware without manual model compilation:

# For 16 GB VRAM systems:
ollama run hf.co/Bigchilli1978/reggie-v6-32b-GGUF:reggie-v6-32b-Q3_K_M.gguf

# For 24 GB+ VRAM systems:
ollama run hf.co/Bigchilli1978/reggie-v6-32b-GGUF:reggie-v6-32b-Q4_K_M.gguf

Or build with a local Modelfile:

FROM ./reggie-v6-32b-Q3_K_M.gguf

TEMPLATE """{{ if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{ if .Prompt }}<|im_start|>user
{{ .Prompt }}<|im_end|>
{{ end }}<|im_start|>assistant
"""

PARAMETER stop "<|im_start|>"
PARAMETER stop "<|im_end|>"
PARAMETER temperature 0.2
PARAMETER top_p 0.95

ollama create reggie -f Modelfile
ollama run reggie

---

🛠️ Highlights & Technical Feats

1. Physical Hardware Sentinels: Directly interfaces with ESP32 microcontrollers, WiFi sniffers, deauth detectors, and raw serial logs.
2. Zero Cloud Quotas / Memory Wall Solution: When cloud GPU/TPU environments exhausted allocated limits, tensor fusion and quantization were achieved on sovereign consumer hardware through custom C++ chunked streaming (llama-export-lora), overcoming the 65 GB full float32 memory wall without CPU paging thrash.
3. Deterministic SCADA & Threat Assessment: Fine-tuned on industrial automation protocols, packet inspection schemas, and network defense policies.