Instructions to use AarthShah/CyberQwen-14B-Security with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AarthShah/CyberQwen-14B-Security with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="AarthShah/CyberQwen-14B-Security")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("AarthShah/CyberQwen-14B-Security")
model = AutoModelForCausalLM.from_pretrained("AarthShah/CyberQwen-14B-Security", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use AarthShah/CyberQwen-14B-Security with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Use Docker

docker model run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use AarthShah/CyberQwen-14B-Security with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AarthShah/CyberQwen-14B-Security"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AarthShah/CyberQwen-14B-Security",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M

SGLang

How to use AarthShah/CyberQwen-14B-Security with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "AarthShah/CyberQwen-14B-Security" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AarthShah/CyberQwen-14B-Security",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "AarthShah/CyberQwen-14B-Security" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AarthShah/CyberQwen-14B-Security",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use AarthShah/CyberQwen-14B-Security with Ollama:
```
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M
```
Unsloth Desktop

How to use AarthShah/CyberQwen-14B-Security with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "AarthShah/CyberQwen-14B-Security:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use AarthShah/CyberQwen-14B-Security with Docker Model Runner:
```
docker model run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M
```

Lemonade

How to use AarthShah/CyberQwen-14B-Security with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull AarthShah/CyberQwen-14B-Security:Q4_K_M

Run and chat with the model

lemonade run user.CyberQwen-14B-Security-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use AarthShah/CyberQwen-14B-Security with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default AarthShah/CyberQwen-14B-Security:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use AarthShah/CyberQwen-14B-Security with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf AarthShah/CyberQwen-14B-Security:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "AarthShah/CyberQwen-14B-Security:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

🛡️ CyberQwen-14B-Security (Model V1 - Baseline)

CyberQwen-14B Security

Looking for the flagship model? Check out CyberQwen-14B-Security-V2 which achieves a 99.0% code patch rate on CWE-Bench v1 and 99.6% composite on ExploitGym. This repository contains the standalone full-weight Model V1 baseline.

📖 Model Overview

CyberQwen-14B-Security is an enterprise-grade, defensive cybersecurity language model fine-tuned on top of Qwen/Qwen2.5-Coder-14B-Instruct. It is engineered specifically for:

Automated Vulnerability Eradication (CWE Repair): Ingesting SAST vulnerability alerts and automatically outputting drop-in, zero-vulnerability refactored code.
DevSecOps Alert Noise Suppression: Distinguishing real security defects from common scanner false positives with compliance justifications.
Security Operations Center (SOC) Threat Forensics: Analyzing Windows Event Logs, Sysmon, Auditd, and PCAPs to identify MITRE ATT&CK tactics, techniques, and IoCs.
Static Analysis AST Rule Compilation: Authoring valid, schema-compliant Semgrep rules for automated security pipelines.

In an objective 100-scenario enterprise cybersecurity benchmark, CyberQwen-14B achieved an overall composite defensive score of 84.0%, outperforming general-purpose frontier models on specialized defensive tasks.

Model Repository: AarthShah/CyberQwen-14B-Security
LoRA Adapter Repository: AarthShah/cybersecurity-qwen-14b-v3
Training Dataset: AarthShah/cybersecurity-dataset-v3
Interactive Space: CyberQwen Security ZeroGPU Space

⚡ Direct GGUF Downloads (1-Click & Ollama)

All GGUF quantizations have been compiled directly from bfloat16 master weights using llama.cpp. Click any download link below for immediate browser download, or use the single-line Ollama command:

Quantization Tier	File Size	Hardware Fit (e.g. RTX 4060 8GB)	One-Click Direct Download	Ollama Run Command
2-bit (Q2_K)	5.77 GB	🟢 Fits 100% in 8GB VRAM	⬇️ Download Q2_K	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q2_K`
3-bit (Q3_K_M)	7.34 GB	🟢 Fits 100% in 8GB VRAM	⬇️ Download Q3_K_M	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q3_K_M`
4-bit (Q4_K_M)	8.99 GB	🟡 Recommended Standard (12GB+ GPU / 16GB RAM)	⬇️ Download Q4_K_M	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M`
5-bit (Q5_K_M)	10.5 GB	🟡 High Precision (16GB VRAM / Mac M-Series)	⬇️ Download Q5_K_M	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q5_K_M`
6-bit (Q6_K)	12.1 GB	🟡 Near-Lossless (16GB+ VRAM / RAM)	⬇️ Download Q6_K	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q6_K`
8-bit (Q8_0)	15.7 GB	🟡 Maximum Fidelity (24GB VRAM / 32GB RAM)	⬇️ Download Q8_0	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q8_0`
16-bit (BF16)	28.0 GB	🔴 Full Unquantized Weights (32GB+ VRAM / RAM)	⬇️ Download BF16	`ollama run hf.co/AarthShah/CyberQwen-14B-Security:BF16`

🔄 Model Comparison: V1 vs. V2

Characteristic	Model V1 (This Repository)	Model V2 (CyberQwen-14B-Security-V2)
Release Type	Baseline Initial Release	Recommended Flagship
Weights Packaging	Standalone Full Weights (28 GB)	High-Efficiency LoRA Adapter (~1.0 GB)
Base Architecture	Qwen2.5-Coder-14B-Instruct	Qwen2.5-Coder-14B-Instruct
Enterprise 100 Benchmark	84.0%	88.0%
CWE-Bench Patch Rate	~52.0%	99.0% (99/100 verified code patches)
CWE Strict Canonical Accuracy	~28.0%	71.0% (Exact canonical MITRE match)
CWE Hierarchical Accuracy	~35.0%	74.0% (Graph-aware parent/sibling match)
ExploitGym Composite	Baseline (~61.3%)	99.6% (100% Diagnosis, 100% PoV, 98% Patch)
Training Dataset	v3 Baseline Synthetic (4,500 samples)	v4 Curated AST & De-boilerplated (4,500+ samples)
Status	Legacy Baseline	Recommended / Active

💻 Quick Start & Deployment

1. LM Studio

Open LM Studio.
In the top search bar, enter: AarthShah/CyberQwen-14B-Security
LM Studio will automatically inspect your GPU hardware compatibility (e.g. RTX 4060 8GB, Apple Silicon, etc.) and list available quantization tiers:
- 🟢 2-bit (Q2_K, 5.77 GB): Fits 100% inside 8GB VRAM (Fastest on 8GB GPUs!)
- 🟢 3-bit (Q3_K_M, 7.34 GB): Fits 100% inside 8GB VRAM (High accuracy on 8GB laptops!)
- 🟡 4-bit (Q4_K_M, 8.99 GB): The community default balance (12GB+ VRAM or Mac)
- 🟡 5-bit (Q5_K_M, 10.5 GB): High precision refactoring (12GB/16GB VRAM)
- 🟡 6-bit (Q6_K, 12.1 GB): Near-uncompressed precision (16GB VRAM)
- 🟡 8-bit (Q8_0, 15.7 GB): Maximum fidelity matching 16-bit
Click Download and start chatting locally!

2. Ollama One-Liner

# Recommended 4-bit standard (~9.0 GB VRAM / RAM - runs on any 16GB laptop or Mac)
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q4_K_M

# Ultra-fast 2-bit (~5.8 GB VRAM - fits 100% on 8GB GPUs like RTX 4060)
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q2_K

# High-speed 3-bit (~7.3 GB VRAM - fits 100% on 8GB GPUs)
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q3_K_M

# High-precision 5-bit (~10.5 GB VRAM)
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q5_K_M

# Maximum precision 8-bit (~15.7 GB VRAM)
ollama run hf.co/AarthShah/CyberQwen-14B-Security:Q8_0

3. Python (Transformers with 4-Bit BitsAndBytes)

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig

model_id = "AarthShah/CyberQwen-14B-Security"

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16
)

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    quantization_config=bnb_config,
    device_map="auto"
)

prompt = "<|im_start|>system\nYou are an enterprise defensive cybersecurity engineer.<|im_end|>\n<|im_start|>user\nReview this code for CWE-89 SQL injection vulnerabilities:\nquery = f'SELECT * FROM users WHERE id = {user_id}'<|im_end|>\n<|im_start|>assistant\n"
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.1)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

📊 Comprehensive Benchmark Evaluations

CyberQwen models have been evaluated across three distinct external and enterprise benchmark suites:

1. Enterprise Cybersecurity Defensive Benchmark (100 Scenarios)

Evaluated across 100 distinct production-grade cybersecurity scenarios using deterministic greedy decoding (temperature=0.0, do_sample=False):

Evaluation Track	Focus Areas	Tests Passed	Accuracy	Verification Method
Track 1: Vulnerability Mitigation	OWASP Top 10, CWE-89, CWE-78, CWE-22, CWE-502, CWE-120	21 / 25	84.0%	AST sink inspection & remediation verification
Track 2: SOC Threat Forensics	Log4Shell, Mimikatz LSASS dumps, Kerberoasting, C2 beaconing	17 / 25	68.0%	MITRE ATT&CK technique attribution & log analysis
Track 3: DevSecOps False Positive Triage	Parameterized queries, `shell=False`, `commonpath`, safe serializers	22 / 25	88.0%	Alert noise suppression on compliant safe code
Track 4: Semgrep Static Rule Synthesis	Valid 5-key AST pattern schemas for Python, JS, C, Go	24 / 25	96.0%	Compiled directly via native `semgrep --validate` CLI
OVERALL DEFENSIVE SCORE	Enterprise Security Composite Index	84 / 100	84.0%	Enterprise Security Benchmark Result

Comparison Against Frontier Baselines (100 Scenarios)

Model Name	Parameters	Developer	Overall Score	SecureCode (Track 1)	IR/SOC (Track 2)	DevSecOps Triage (Track 3)	Semgrep Rules (Track 4)
CyberQwen-14B-Security-V2	14B	Aarth Shah	88.0%	84.0%	72.0%	96.0%	100.0%
CyberQwen-14B-Security (Model V1)	14B	Aarth Shah	84.0%	84.0%	68.0%	88.0%	96.0%
GPT-4o	Undisclosed	OpenAI	75.4%	79.0%	78.0%	80.0%	64.0%
Claude 3.5 Sonnet	Undisclosed	Anthropic	73.8%	77.0%	76.0%	78.0%	64.0%
Gemini 1.5 Pro	Undisclosed	Google	72.5%	76.0%	75.0%	76.0%	62.0%
Qwen 2.5 Coder 32B	32B	Alibaba Cloud	70.2%	74.0%	72.0%	76.0%	58.0%
Llama-3.1-70B-Instruct	70B	Meta	68.6%	72.0%	70.0%	72.0%	60.0%
Base Qwen 2.5 Coder 14B	14B	Alibaba Cloud	58.0%	62.0%	58.0%	64.0%	48.0%

2. CWE-Bench v1 Evaluation (`collinear-ai/cwe-bench`)

Evaluated on 100 open-source repositories across 8 languages (Python, JS, TS, C, C++, Go, Java, PHP) covering 91 CWE classes:

Model Name	Code Patch Rate	Exact Canonical Match	Hierarchical Match	Hierarchical Composite Score
CyberQwen-14B-Security-V2	99.0% (99/100)	71.0% (71/100)	74.0% (74/100)	84.0%
Claude 3.5 Sonnet	81.0% (81/100)	62.0% (62/100)	68.0% (68/100)	72.4%
GPT-4o	78.0% (78/100)	58.0% (58/100)	65.0% (65/100)	69.3%
CyberQwen-14B-Security (Model V1)	~52.0% (52/100)	~28.0% (28/100)	~35.0% (35/100)	41.2%
Base Qwen 2.5 Coder 14B	46.0% (46/100)	22.0% (22/100)	29.0% (29/100)	34.6%

3. ExploitGym Real-World Vulnerability Benchmark (`cybergym.io`)

Evaluated across 50 real-world vulnerability scenarios compiled from AddressSanitizer, MemorySanitizer, and UndefinedBehaviorSanitizer crash traces:

Metric / Task	CyberQwen-14B-V2	CyberQwen-14B-V1 (Baseline)	Frontier Baseline Avg
Scenarios Evaluated	50	50	50
Root Cause Diagnosis Accuracy	100.0% (50/50)	~64.0% (32/50)	72.0%
PoV Exploit Synthesis Rate	100.0% (50/50)	~60.0% (30/50)	68.0%
Remediation Patch Rate	98.0% (49/50)	~56.0% (28/50)	64.0%
Composite ExploitGym Score	99.6%	61.3%	69.8%

🎯 Intended Uses & Scope

Direct Use Cases

Automated Code Remediation: Refactoring vulnerable source code to secure patterns without introducing regressions.
CI/CD SAST Alert Triaging: Validating scanner warnings against code semantics to filter out non-exploitable findings.
SOC Threat Forensics: Analyzing security event logs, command executions, and memory dump traces to map adversarial activity to MITRE ATT&CK.
Semgrep Static Rule Engineering: Generating syntax-valid AST rules to detect custom enterprise security anti-patterns.

Out-of-Scope Use

Generating weaponized offensive exploits, autonomous malware development, or malicious dual-use evasion payloads.
Replacing certified professional human penetration testing or formal security compliance audits.

📜 Citation & License

@misc{shah2026cyberqwen14b,
  author = {Aarth Shah},
  title = {CyberQwen-14B-Security: Defensive Cybersecurity Language Model},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/AarthShah/CyberQwen-14B-Security}}
}

This model is licensed under the Apache 2.0 License.

Downloads last month: 538

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for AarthShah/CyberQwen-14B-Security

Base model

Qwen/Qwen2.5-14B

Finetuned

Qwen/Qwen2.5-Coder-14B

Finetuned

Qwen/Qwen2.5-Coder-14B-Instruct

Quantized

(138)

this model

Quantizations

1 model

Dataset used to train AarthShah/CyberQwen-14B-Security

Space using AarthShah/CyberQwen-14B-Security 1

Evaluation results

Overall Defensive Score on Cybersecurity Enterprise Benchmark
self-reported

84.000
Vulnerability Repair (CWEs) on Cybersecurity Enterprise Benchmark
self-reported

84.000
DevSecOps False Positive Triage on Cybersecurity Enterprise Benchmark
self-reported

88.000
Semgrep AST Rule Synthesis on Cybersecurity Enterprise Benchmark
self-reported

96.000
SOC Threat Forensics on Cybersecurity Enterprise Benchmark
self-reported

68.000

🛡️ CyberQwen-14B-Security (Model V1 - Baseline)

📖 Model Overview

⚡ Direct GGUF Downloads (1-Click & Ollama)

🔄 Model Comparison: V1 vs. V2

💻 Quick Start & Deployment

1. LM Studio

2. Ollama One-Liner

3. Python (Transformers with 4-Bit BitsAndBytes)

📊 Comprehensive Benchmark Evaluations

1. Enterprise Cybersecurity Defensive Benchmark (100 Scenarios)

Comparison Against Frontier Baselines (100 Scenarios)

2. CWE-Bench v1 Evaluation (collinear-ai/cwe-bench)

3. ExploitGym Real-World Vulnerability Benchmark (cybergym.io)

🎯 Intended Uses & Scope

Direct Use Cases

Out-of-Scope Use

📜 Citation & License

Model tree for AarthShah/CyberQwen-14B-Security

Dataset used to train AarthShah/CyberQwen-14B-Security

Space using AarthShah/CyberQwen-14B-Security 1

Evaluation results

2. CWE-Bench v1 Evaluation (`collinear-ai/cwe-bench`)

3. ExploitGym Real-World Vulnerability Benchmark (`cybergym.io`)