---
license: cc0-1.0
base_model: mlx-community/Qwen2.5-Coder-7B-Instruct-4bit
tags:
- gguf
- cybersecurity
- nist
- security-controls
- compliance
- fine-tuned
- llama-cpp
language:
- en
quantized_by: ethanolivertroy
---

# HackIDLE-NIST-Coder v1.1 (GGUF)

**The most comprehensive NIST cybersecurity model** in GGUF format - Compatible with llama.cpp, Ollama, LM Studio, and text-generation-webui.

## Model Overview

Fine-tuned on 530,912 examples from 596 NIST publications. Version 1.1 includes:

- **+7,206 training examples** (530,912 total)
- **+28 new documents** (596 NIST publications)
- **CSWP series**: CSF 2.0, Zero Trust Architecture, Post-Quantum Cryptography
- **Improved quality**: Fixed 6,150 malformed DOI links, 0 broken link markers

## Available Quantizations

| Quantization | Size | Use Case | Description |
|--------------|------|----------|-------------|
| **F16** | ~14 GB | Reference Quality | Full precision, best quality |
| **Q8_0** | ~7.5 GB | High Quality | Minimal quality loss |
| **Q5_K_M** | ~5.1 GB | Balanced | Good quality/size trade-off |
| **Q4_K_M** | ~4.4 GB | Recommended | Best speed/quality balance |

**Recommended**: Start with **Q4_K_M** for best overall performance.

## Training Data (v1.1)

**Dataset**: [ethanolivertroy/nist-cybersecurity-training](https://huggingface.co/datasets/ethanolivertroy/nist-cybersecurity-training)

**Coverage:**
- **FIPS**: Cryptographic standards
- **SP 800**: Security guidelines and controls
- **SP 1800**: Practice guides
- **IR**: Technical reports
- **CSWP**: White Papers (CSF 2.0, Zero Trust, PQC, IoT, Privacy) ✨ NEW

**Stats**: 530,912 examples • 596 documents • 61,480 working references

## Installation

### Ollama

```bash
# Pull from Ollama registry
ollama pull etgohome/hackidle-nist-coder:v1.1

# Or create from GGUF
ollama create hackidle-nist-coder -f Modelfile
```

### LM Studio

1. Open LM Studio
2. Search for "hackidle-nist-coder"
3. Download Q4_K_M or Q5_K_M quantization
4. Load and chat

### llama.cpp

```bash
# Clone llama.cpp
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp && make

# Download model (Q4_K_M recommended)
wget https://huggingface.co/ethanolivertroy/HackIDLE-NIST-Coder-v1.1-GGUF/resolve/main/hackidle-nist-coder-v1.1-q4_k_m.gguf

# Run inference
./llama-cli -m hackidle-nist-coder-v1.1-q4_k_m.gguf -p "What is Zero Trust Architecture?"
```

### text-generation-webui

1. Place GGUF file in `models/` directory
2. Select model in UI
3. Load and chat

## Usage Examples

### Ollama

```bash
ollama run etgohome/hackidle-nist-coder:v1.1 "Explain the CSF 2.0 GOVERN function"
```

### Python (llama-cpp-python)

```python
from llama_cpp import Llama

llm = Llama(
    model_path="hackidle-nist-coder-v1.1-q4_k_m.gguf",
    n_ctx=4096,
    n_threads=8
)

response = llm("What are the core principles of Zero Trust Architecture in SP 800-207?",
               max_tokens=500)
print(response['choices'][0]['text'])
```

## Model Capabilities

Trained on comprehensive NIST content:

✅ **Security Controls** (SP 800-53)
✅ **CSF 2.0** with GOVERN function
✅ **Zero Trust Architecture** (SP 800-207)
✅ **Risk Management Framework** (RMF)
✅ **Cloud Security** (SP 800-145, 800-146)
✅ **FIPS Cryptography** standards
✅ **Post-Quantum Cryptography** migration
✅ **Privacy Engineering**
✅ **Supply Chain Risk Management**
✅ **IoT Cybersecurity**

## What's New in v1.1

**Added Content:**
- CSF 2.0 (Cybersecurity Framework 2.0)
- Zero Trust Architecture planning guidance
- Post-Quantum Cryptography recommendations
- IoT security and labeling
- Privacy Framework v1.0
- Supply chain risk management case studies

**Quality Improvements:**
- Fixed 6,150 malformed DOI links
- Removed 202 broken link markers
- Validated 124,946 total links
- Clean training data

## System Requirements

| Quantization | RAM Required | CPU/GPU |
|--------------|-------------|---------|
| Q4_K_M | 6 GB | CPU or GPU |
| Q5_K_M | 7 GB | CPU or GPU |
| Q8_0 | 10 GB | CPU or GPU |
| F16 | 16 GB | GPU recommended |

## Other Formats

- **MLX**: [ethanolivertroy/HackIDLE-NIST-Coder-v1.1-MLX-4bit](https://huggingface.co/ethanolivertroy/HackIDLE-NIST-Coder-v1.1-MLX-4bit) (Apple Silicon)
- **Ollama**: [etgohome/hackidle-nist-coder](https://ollama.com/etgohome/hackidle-nist-coder)

## Limitations

- Training data current as of October 2025
- May not reflect NIST publications released after training
- 54.2% of references are broken links (cataloged for recovery)
- Optimized for NIST-specific cybersecurity questions

## Citation

```bibtex
@misc{hackidle-nist-coder-v1.1-gguf,
  title={HackIDLE-NIST-Coder: NIST Cybersecurity Expert Model},
  author={Troy, Ethan Oliver},
  year={2025},
  version={1.1},
  format={GGUF},
  url={https://huggingface.co/ethanolivertroy/HackIDLE-NIST-Coder-v1.1-GGUF}
}
```

## License

CC0 1.0 Universal (Public Domain) - All NIST publications are in the public domain.

## Acknowledgments

- NIST Computer Security Resource Center
- Qwen2.5-Coder base model (Alibaba Cloud)
- llama.cpp quantization (Georgi Gerganov)
- MLX framework (Apple)

---

**Version**: 1.1
**Release Date**: October 2025
**Training Dataset**: [nist-cybersecurity-training v1.1](https://huggingface.co/datasets/ethanolivertroy/nist-cybersecurity-training)
**Format**: GGUF (compatible with llama.cpp ecosystem)