Text Generation
Transformers
Safetensors
GGUF
English
cybersecurity
soc
siem
mitre-attack
incident-response
threat-detection
security-operations
fine-tuned
qlora
unsloth
ollama
Eval Results (legacy)
conversational
Instructions to use SyedCode01/rhythmai-cybersec-20b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use SyedCode01/rhythmai-cybersec-20b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="SyedCode01/rhythmai-cybersec-20b") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("SyedCode01/rhythmai-cybersec-20b", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use SyedCode01/rhythmai-cybersec-20b 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 SyedCode01/rhythmai-cybersec-20b:MXFP4 # Run inference directly in the terminal: llama cli -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama serve -hf SyedCode01/rhythmai-cybersec-20b:MXFP4 # Run inference directly in the terminal: llama cli -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
Use pre-built binary
# Download pre-built binary from: # https://github.com/ggerganov/llama.cpp/releases # Start a local OpenAI-compatible server with a web UI: ./llama-server -hf SyedCode01/rhythmai-cybersec-20b:MXFP4 # Run inference directly in the terminal: ./llama-cli -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
Build from source code
git clone https://github.com/ggerganov/llama.cpp.git cd llama.cpp cmake -B build cmake --build build -j --target llama-server llama-cli # Start a local OpenAI-compatible server with a web UI: ./build/bin/llama-server -hf SyedCode01/rhythmai-cybersec-20b:MXFP4 # Run inference directly in the terminal: ./build/bin/llama-cli -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
Use Docker
docker model run hf.co/SyedCode01/rhythmai-cybersec-20b:MXFP4
- LM Studio
- Jan
- vLLM
How to use SyedCode01/rhythmai-cybersec-20b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "SyedCode01/rhythmai-cybersec-20b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "SyedCode01/rhythmai-cybersec-20b", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/SyedCode01/rhythmai-cybersec-20b:MXFP4
- SGLang
How to use SyedCode01/rhythmai-cybersec-20b 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 "SyedCode01/rhythmai-cybersec-20b" \ --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": "SyedCode01/rhythmai-cybersec-20b", "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 "SyedCode01/rhythmai-cybersec-20b" \ --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": "SyedCode01/rhythmai-cybersec-20b", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Ollama
How to use SyedCode01/rhythmai-cybersec-20b with Ollama:
ollama run hf.co/SyedCode01/rhythmai-cybersec-20b:MXFP4
- Unsloth Studio
How to use SyedCode01/rhythmai-cybersec-20b with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for SyedCode01/rhythmai-cybersec-20b to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for SyedCode01/rhythmai-cybersec-20b to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for SyedCode01/rhythmai-cybersec-20b to start chatting
- Pi
How to use SyedCode01/rhythmai-cybersec-20b with Pi:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
Configure the model in Pi
# Install Pi: npm install -g @mariozechner/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "llama-cpp": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "SyedCode01/rhythmai-cybersec-20b:MXFP4" } ] } } }Run Pi
# Start Pi in your project directory: pi
- OpenClaw new
How to use SyedCode01/rhythmai-cybersec-20b with OpenClaw:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
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 "SyedCode01/rhythmai-cybersec-20b:MXFP4" \ --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"
- Docker Model Runner
How to use SyedCode01/rhythmai-cybersec-20b with Docker Model Runner:
docker model run hf.co/SyedCode01/rhythmai-cybersec-20b:MXFP4
- Lemonade
How to use SyedCode01/rhythmai-cybersec-20b with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull SyedCode01/rhythmai-cybersec-20b:MXFP4
Run and chat with the model
lemonade run user.rhythmai-cybersec-20b-MXFP4
List all available models
lemonade list
- Hermes Agent
How to use SyedCode01/rhythmai-cybersec-20b with Hermes Agent:
Start the llama.cpp server
# Install llama.cpp: brew install llama.cpp # Start a local OpenAI-compatible server: llama serve -hf SyedCode01/rhythmai-cybersec-20b:MXFP4
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 SyedCode01/rhythmai-cybersec-20b:MXFP4
Run Hermes
hermes
- Atomic Chat
Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -60,14 +60,14 @@ Built for [RhythmAI](https://github.com/SyedCode01) -- an AI-powered SOC platfor
|
|
| 60 |
|
| 61 |
## Training Data
|
| 62 |
|
| 63 |
-
Fine-tuned on **9,702 curated cybersecurity examples** from public datasets, filtered for SOC/SIEM relevance:
|
| 64 |
|
| 65 |
-
| Source |
|
| 66 |
-
|--------|----------
|
| 67 |
| [Fenrir v2.0](https://huggingface.co/datasets/AlicanKiraz0/Cybersecurity-Dataset-Fenrir-v2.0) | 83,920 | ~5,000 | General cybersecurity Q&A |
|
| 68 |
| [Trendyol Cybersecurity](https://huggingface.co/datasets/Trendyol/Trendyol-Cybersecurity-Instruction-Tuning-Dataset) | 53,202 | ~5,000 | Instruction-tuned cybersecurity |
|
| 69 |
|
| 70 |
-
**Filtering
|
| 71 |
|
| 72 |
**Split**: 9,217 train (95%) / 485 validation (5%)
|
| 73 |
|
|
@@ -76,6 +76,84 @@ Fine-tuned on **9,702 curated cybersecurity examples** from public datasets, fil
|
|
| 76 |
{"messages": [{"role": "system", "content": "..."}, {"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]}
|
| 77 |
```
|
| 78 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 79 |
## Training Details
|
| 80 |
|
| 81 |
| Parameter | Value |
|
|
|
|
| 60 |
|
| 61 |
## Training Data
|
| 62 |
|
| 63 |
+
Fine-tuned on **9,702 curated cybersecurity examples** sourced from **137,122 raw examples** across 4 public datasets, aggressively filtered for SOC/SIEM relevance (7.1% acceptance rate):
|
| 64 |
|
| 65 |
+
| Source | Raw Size | After Filtering | Description |
|
| 66 |
+
|--------|----------|-----------------|-------------|
|
| 67 |
| [Fenrir v2.0](https://huggingface.co/datasets/AlicanKiraz0/Cybersecurity-Dataset-Fenrir-v2.0) | 83,920 | ~5,000 | General cybersecurity Q&A |
|
| 68 |
| [Trendyol Cybersecurity](https://huggingface.co/datasets/Trendyol/Trendyol-Cybersecurity-Instruction-Tuning-Dataset) | 53,202 | ~5,000 | Instruction-tuned cybersecurity |
|
| 69 |
|
| 70 |
+
**Filtering pipeline**: Keyword relevance scoring (minimum 2 matches from 60+ SOC-relevant terms), response length between 50-15,000 characters, MD5-based deduplication. Average response length: **2,627 characters (~656 tokens)**.
|
| 71 |
|
| 72 |
**Split**: 9,217 train (95%) / 485 validation (5%)
|
| 73 |
|
|
|
|
| 76 |
{"messages": [{"role": "system", "content": "..."}, {"role": "user", "content": "..."}, {"role": "assistant", "content": "..."}]}
|
| 77 |
```
|
| 78 |
|
| 79 |
+
### Cybersecurity Content Breakdown
|
| 80 |
+
|
| 81 |
+
#### MITRE ATT&CK Coverage
|
| 82 |
+
|
| 83 |
+
The training data references **424 unique MITRE ATT&CK technique IDs** across all 14 tactics:
|
| 84 |
+
|
| 85 |
+
| Tactic | Examples | Coverage |
|
| 86 |
+
|--------|----------|----------|
|
| 87 |
+
| Execution | 3,004 | 31.0% |
|
| 88 |
+
| Lateral Movement | 2,427 | 25.0% |
|
| 89 |
+
| Impact | 1,949 | 20.1% |
|
| 90 |
+
| Privilege Escalation | 1,637 | 16.9% |
|
| 91 |
+
| Persistence | 1,568 | 16.2% |
|
| 92 |
+
| Exfiltration | 1,425 | 14.7% |
|
| 93 |
+
| Defense Evasion | 1,277 | 13.2% |
|
| 94 |
+
| Collection | 1,080 | 11.1% |
|
| 95 |
+
| Reconnaissance | 900 | 9.3% |
|
| 96 |
+
| Discovery | 889 | 9.2% |
|
| 97 |
+
| Initial Access | 807 | 8.3% |
|
| 98 |
+
| Command and Control | 208 | 2.1% |
|
| 99 |
+
| Credential Access | 169 | 1.7% |
|
| 100 |
+
| Resource Development | 12 | 0.1% |
|
| 101 |
+
|
| 102 |
+
**Most referenced techniques**: T1078 (Valid Accounts, 1,451 examples), T1055 (Process Injection, 1,120), T1021 (Remote Services, 582), T1071 (Application Layer Protocol, 541), T1027 (Obfuscated Files, 378), T1566 (Phishing, 378), T1059 (Command and Scripting Interpreter, 376), T1562 (Impair Defenses, 339), T1203 (Exploitation for Client Execution, 323), T1041 (Exfiltration Over C2, 322).
|
| 103 |
+
|
| 104 |
+
#### Attack Types & Threat Categories
|
| 105 |
+
|
| 106 |
+
| Attack Type | Examples | Coverage |
|
| 107 |
+
|-------------|----------|----------|
|
| 108 |
+
| Phishing & Social Engineering | 9,546 | 98.4% |
|
| 109 |
+
| Remote Code Execution | 5,620 | 57.9% |
|
| 110 |
+
| Lateral Movement | 2,427 | 25.0% |
|
| 111 |
+
| Privilege Escalation | 1,637 | 16.9% |
|
| 112 |
+
| PowerShell-based Attacks | 731 | 7.5% |
|
| 113 |
+
| Supply Chain Attacks | 653 | 6.7% |
|
| 114 |
+
| Credential Dumping (Mimikatz/LSASS) | 393 | 4.1% |
|
| 115 |
+
| Insider Threats | 376 | 3.9% |
|
| 116 |
+
| Zero-Day Exploits | 375 | 3.9% |
|
| 117 |
+
| Man-in-the-Middle | 294 | 3.0% |
|
| 118 |
+
| Brute Force / Credential Stuffing | 264 | 2.7% |
|
| 119 |
+
| C2 Communication | 228 | 2.4% |
|
| 120 |
+
| DDoS / Denial of Service | 217 | 2.2% |
|
| 121 |
+
| Backdoors | 203 | 2.1% |
|
| 122 |
+
| Rootkits | 180 | 1.9% |
|
| 123 |
+
| SQL Injection | 177 | 1.8% |
|
| 124 |
+
| Buffer Overflow | 144 | 1.5% |
|
| 125 |
+
| Cross-Site Scripting (XSS) | 127 | 1.3% |
|
| 126 |
+
| Fileless Malware | 116 | 1.2% |
|
| 127 |
+
| Living Off The Land (LOLBins) | 80 | 0.8% |
|
| 128 |
+
| DNS Tunneling | 57 | 0.6% |
|
| 129 |
+
|
| 130 |
+
#### Log Source & SIEM Knowledge
|
| 131 |
+
|
| 132 |
+
| Log Type | Examples | Coverage |
|
| 133 |
+
|----------|----------|----------|
|
| 134 |
+
| Windows Event Logs (Event IDs) | 977 | 10.1% |
|
| 135 |
+
| Network Flow (NetFlow/PCAP) | 410 | 4.2% |
|
| 136 |
+
| IDS/IPS Alerts | 364 | 3.8% |
|
| 137 |
+
| Authentication Logs | 289 | 3.0% |
|
| 138 |
+
| Firewall Logs | 150 | 1.5% |
|
| 139 |
+
| DNS Logs | 123 | 1.3% |
|
| 140 |
+
| Syslog | 112 | 1.2% |
|
| 141 |
+
|
| 142 |
+
**Security platforms referenced**: Nmap (214), YARA rules (158), Microsoft Sentinel (120), Elastic/ELK (107), Wireshark (104), Splunk (70), Metasploit (65), Sigma rules (50), Snort/Suricata (45).
|
| 143 |
+
|
| 144 |
+
#### Compliance & Regulatory Frameworks
|
| 145 |
+
|
| 146 |
+
| Framework | Examples | Coverage |
|
| 147 |
+
|-----------|----------|----------|
|
| 148 |
+
| NIST (CSF/SP 800-series) | 9,620 | 99.2% |
|
| 149 |
+
| GDPR | 411 | 4.2% |
|
| 150 |
+
| HIPAA | 310 | 3.2% |
|
| 151 |
+
| OWASP | 304 | 3.1% |
|
| 152 |
+
| PCI-DSS | 152 | 1.6% |
|
| 153 |
+
| CIS Controls | 66 | 0.7% |
|
| 154 |
+
| ISO 27001 | 57 | 0.6% |
|
| 155 |
+
| SOC 2 | 35 | 0.4% |
|
| 156 |
+
|
| 157 |
## Training Details
|
| 158 |
|
| 159 |
| Parameter | Value |
|