README.md

---
license: apache-2.0
language:
- en
library_name: transformers
tags:
- quantum-computing
- post-quantum-cryptography
- security
- pqc
- nist
- llm
- gguf
- ollama
- qwen2
base_model: Qwen/Qwen2.5-3B-Instruct
pipeline_tag: text-generation
datasets:
- custom
model-index:
- name: quantum-readiness-gui
  results: []
---

# Quantum Readiness LLM (GUI Version)

A fine-tuned language model specialized in **quantum computing readiness assessment**, including post-quantum cryptography (PQC) vulnerability scanning and quantum feasibility analysis.

## Model Description

This model is fine-tuned from [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct) using QLoRA (4-bit quantization) on a custom dataset of quantum security and quantum computing analysis examples.

### Key Features

- **Post-Quantum Cryptography (PQC) Analysis**: Identifies quantum-vulnerable algorithms (RSA, ECDSA, ECDH, DSA, DH, DES, 3DES, RC4, MD5, SHA-1)
- **NIST Compliance Guidance**: References FIPS 203 (ML-KEM/Kyber), FIPS 204 (ML-DSA/Dilithium), FIPS 205 (SLH-DSA/SPHINCS+)
- **Quantum Feasibility Analysis (QFA)**: Identifies opportunities for quantum algorithms (QAOA, VQE, Grover's, HHL, QML)
- **Migration Guidance**: Provides code examples for transitioning to post-quantum alternatives
- **Standalone Operation**: Provides direct responses without requiring tool orchestration

## Model Details

| Property | Value |
|----------|-------|
| Base Model | Qwen/Qwen2.5-3B-Instruct |
| Parameters | ~3B |
| Fine-tuning Method | QLoRA (4-bit NF4) |
| LoRA Rank | 32 |
| LoRA Alpha | 64 |
| Training Epochs | 3 |
| Final Loss | 0.02 |
| Token Accuracy | 99% |
| Context Length | 4096 tokens |
| Format | GGUF (F16) |
| Size | ~6.2 GB |

## Intended Use

### Primary Use Cases

1. **Security Auditing**: Scan codebases for quantum-vulnerable cryptographic implementations
2. **Compliance Assessment**: Evaluate alignment with NIST post-quantum standards
3. **Migration Planning**: Get guidance on transitioning to post-quantum algorithms
4. **Quantum Opportunity Analysis**: Identify code patterns suitable for quantum acceleration
5. **Education**: Learn about quantum threats and post-quantum cryptography

### Example Prompts

```
# PQC Security Scanning
"What cryptographic algorithms are vulnerable to quantum attacks?"
"How do I replace RSA with post-quantum alternatives?"
"Explain the quantum threat to ECDSA"

# Quantum Feasibility
"What quantum algorithms could benefit my optimization problems?"
"Explain QAOA and its use cases"
"What is the Quantum Feasibility Index?"

# NIST Standards
"What are the NIST PQC standards?"
"Explain ML-KEM (Kyber) and FIPS 203"
"What's the difference between ML-DSA and SLH-DSA?"
```

## How to Use

### With Ollama (Recommended)

1. Download the GGUF file and Modelfile
2. Create the model:
```bash
ollama create quantum-readiness-gui -f Modelfile
```
3. Run:
```bash
ollama run quantum-readiness-gui "What algorithms are vulnerable to quantum attacks?"
```

### Modelfile Configuration

```dockerfile
FROM quantum-readiness-gui-f16.gguf

PARAMETER num_ctx 4096
PARAMETER temperature 0.7
PARAMETER top_p 0.9
PARAMETER repeat_penalty 1.1

SYSTEM """You are the ArcQubit Quantum Assistant, an AI-powered expert in quantum computing readiness, post-quantum cryptography (PQC), and quantum feasibility analysis."""

TEMPLATE """{{- if .System }}<|im_start|>system
{{ .System }}<|im_end|>
{{ end }}{{- range .Messages }}<|im_start|>{{ .Role }}
{{ .Content }}<|im_end|>
{{ end }}<|im_start|>assistant
"""
```

### With llama.cpp

```bash
./main -m quantum-readiness-gui-f16.gguf \
  -p "<|im_start|>user\nWhat algorithms are vulnerable to quantum attacks?<|im_end|>\n<|im_start|>assistant\n" \
  -n 512 --temp 0.7
```

### With Transformers (Merged Model)

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("path/to/merged-model-gui")
tokenizer = AutoTokenizer.from_pretrained("path/to/merged-model-gui")

messages = [
    {"role": "user", "content": "What is ML-KEM (Kyber)?"}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=512)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

## Training Data

The model was trained on ~490 examples covering:

- **PQC Vulnerability Scanning**: RSA, ECDSA, ECDH, DSA, DH, DES, 3DES, RC4, MD5, SHA-1 detection and remediation
- **Quantum Feasibility Analysis**: QAOA, VQE, Grover's, HHL, QML opportunity identification
- **Migration Guides**: Step-by-step code examples for transitioning to post-quantum algorithms
- **NIST Standards**: FIPS 203, 204, 205 compliance guidance

### Training Configuration

```yaml
model:
  base_model: "Qwen/Qwen2.5-3B-Instruct"
  load_in_4bit: true
  bnb_4bit_quant_type: "nf4"

lora:
  r: 32
  lora_alpha: 64
  lora_dropout: 0.05
  target_modules: ["q_proj", "k_proj", "v_proj", "o_proj", "gate_proj", "up_proj", "down_proj"]

training:
  num_train_epochs: 3
  per_device_train_batch_size: 4
  gradient_accumulation_steps: 4
  learning_rate: 2.0e-4
  lr_scheduler_type: "cosine"
```

## NIST PQC Standards Reference

| Algorithm | Purpose | Standard | Status |
|-----------|---------|----------|--------|
| ML-KEM (Kyber) | Key Encapsulation | FIPS 203 | Finalized |
| ML-DSA (Dilithium) | Digital Signatures | FIPS 204 | Finalized |
| SLH-DSA (SPHINCS+) | Stateless Signatures | FIPS 205 | Finalized |

## Quantum Algorithm Categories

| Category | Algorithm | Use Case |
|----------|-----------|----------|
| Optimization | QAOA | Portfolio optimization, scheduling, logistics |
| Simulation | VQE | Chemistry, materials science |
| Search | Grover's | Database search, cryptanalysis |
| Linear Algebra | HHL | Machine learning, optimization |
| Machine Learning | QML | Classification, clustering |

## Limitations

- **Not a Security Tool**: This model provides guidance but should not replace professional security audits
- **General Patterns**: File paths and line numbers in responses are illustrative, not from actual scans
- **Knowledge Cutoff**: Based on NIST standards as of early 2025
- **No Code Execution**: The model provides analysis but does not execute code or access repositories

## Ethical Considerations

- Use responsibly for security improvement, not for identifying vulnerabilities to exploit
- Verify all recommendations with security professionals before implementation
- The model is intended for defensive security purposes only

## Files

| File | Description | Size |
|------|-------------|------|
| `quantum-readiness-gui-f16.gguf` | GGUF model (F16 quantization) | ~6.2 GB |
| `Modelfile` | Ollama configuration | ~2 KB |

## Citation

```bibtex
@misc{quantum-readiness-llm,
  title={Quantum Readiness LLM: A Fine-tuned Model for Post-Quantum Cryptography Analysis},
  author={ArcQubit},
  year={2024},
  publisher={Hugging Face},
  url={https://huggingface.co/tbowman/quantum-readiness-gui}
}
```

## License

Apache 2.0

## Acknowledgments

- Base model: [Qwen/Qwen2.5-3B-Instruct](https://huggingface.co/Qwen/Qwen2.5-3B-Instruct)
- NIST Post-Quantum Cryptography Standardization
- Hugging Face Transformers and PEFT libraries