GGUF
English
llama
unsloth
llama-3.2
3b
cybersecurity
instruction-tuning
conversational-ai
penetration-testing
chain-of-thought
ollama
conversational
Instructions to use saberbx/XO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- llama-cpp-python
How to use saberbx/XO with llama-cpp-python:
# !pip install llama-cpp-python from llama_cpp import Llama llm = Llama.from_pretrained( repo_id="saberbx/XO", filename="unsloth.F16.gguf", )
llm.create_chat_completion( messages = "No input example has been defined for this model task." )
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- llama.cpp
How to use saberbx/XO with llama.cpp:
Install from brew
brew install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama-server -hf saberbx/XO:F16 # Run inference directly in the terminal: llama-cli -hf saberbx/XO:F16
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama-server -hf saberbx/XO:F16 # Run inference directly in the terminal: llama-cli -hf saberbx/XO:F16
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 saberbx/XO:F16 # Run inference directly in the terminal: ./llama-cli -hf saberbx/XO:F16
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 saberbx/XO:F16 # Run inference directly in the terminal: ./build/bin/llama-cli -hf saberbx/XO:F16
Use Docker
docker model run hf.co/saberbx/XO:F16
- LM Studio
- Jan
- Ollama
How to use saberbx/XO with Ollama:
ollama run hf.co/saberbx/XO:F16
- Unsloth Studio
How to use saberbx/XO 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 saberbx/XO 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 saberbx/XO to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for saberbx/XO to start chatting
- Docker Model Runner
How to use saberbx/XO with Docker Model Runner:
docker model run hf.co/saberbx/XO:F16
- Lemonade
How to use saberbx/XO with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull saberbx/XO:F16
Run and chat with the model
lemonade run user.XO-F16
List all available models
lemonade list
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-sa-4.0
|
| 3 |
+
language:
|
| 4 |
+
- en
|
| 5 |
+
tags:
|
| 6 |
+
- unsloth
|
| 7 |
+
- llama-3.2
|
| 8 |
+
- 3b
|
| 9 |
+
- cybersecurity
|
| 10 |
+
- instruction-tuning
|
| 11 |
+
- conversational-ai
|
| 12 |
+
- penetration-testing
|
| 13 |
+
- chain-of-thought
|
| 14 |
+
- gguf
|
| 15 |
+
- ollama
|
| 16 |
+
base_model: unsloth/llama-3.2-3b-instruct-nb-bnb-4bit
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+

|
| 20 |
+
|
| 21 |
+
# XO: A Llama 3.2 3B, Unsloth-Trained Cybersecurity Expert
|
| 22 |
+
|
| 23 |
+
## Model Description
|
| 24 |
+
|
| 25 |
+
**XO** is an instruction-fine-tuned model based on **`unsloth/llama-3.2-3b-instruct-nb-bnb-4bit`**. It is engineered to be a lightweight, efficient, and highly specialized AI assistant for cybersecurity tasks. Its small size makes it ideal for local deployment on consumer-grade hardware using tools like Ollama or LM Studio.
|
| 26 |
+
|
| 27 |
+
The model was fine-tuned using the **Unsloth** framework, ensuring maximum performance and minimal resource consumption from the 3B parameter architecture. This version of XO is trained on a focused, foundational dataset to provide core cybersecurity knowledge and a consistent persona in English.
|
| 28 |
+
|
| 29 |
+
## Model Details
|
| 30 |
+
|
| 31 |
+
* **Model Type:** Fine-tuned Causal Language Model
|
| 32 |
+
* **Base Model:** [unsloth/llama-3.2-3b-instruct-nb-bnb-4bit](https://huggingface.co/unsloth/llama-3.2-3b-instruct-nb-bnb-4bit)
|
| 33 |
+
* **Training Framework:** [Unsloth](https://github.com/unslothai/unsloth) 🚀
|
| 34 |
+
* **Training Data:** The model was trained on the foundational, English-only **[`saberbx/X-mini-datasets`](https://huggingface.co/datasets/saberbx/X-mini-datasets)**. This dataset includes:
|
| 35 |
+
* Core knowledge adapted from the "Payloads All The Things" repository.
|
| 36 |
+
* An introductory Chain-of-Thought module for basic reasoning.
|
| 37 |
+
* A persona module to define its identity as "XO," created by "Saber."
|
| 38 |
+
* **Important Note:** This model is **NOT** trained on the advanced, bilingual dataset and does **NOT** include advanced mathematical reasoning capabilities.
|
| 39 |
+
|
| 40 |
+
## Capabilities & Intended Use
|
| 41 |
+
|
| 42 |
+
XO is designed to be a reliable local assistant for day-to-day cybersecurity tasks. Its primary capabilities include:
|
| 43 |
+
|
| 44 |
+
* 💻 **Optimized for Local Deployment:** Its 3B parameter size allows it to run smoothly on machines with limited VRAM, making powerful AI accessible.
|
| 45 |
+
* 🛡️ **Core Cybersecurity Knowledge:** Acts as an interactive encyclopedia of "Payloads All The Things," providing quick access to common payloads, commands, and checklists.
|
| 46 |
+
* 🧠 **Foundational Reasoning:** Capable of performing basic step-by-step analysis for common cybersecurity problems based on its Chain-of-Thought training.
|
| 47 |
+
* 👤 **Consistent Persona:** Always responds as "XO," the AI assistant created by "Saber," providing a consistent and predictable user experience.
|
| 48 |
+
|
| 49 |
+
## Limitations and Ethical Considerations
|
| 50 |
+
|
| 51 |
+
* **⚠️ For Ethical & Defensive Use Only:** This model is designed to empower cybersecurity professionals. **Any use for malicious or illegal activities is strictly prohibited.**
|
| 52 |
+
* **Limited Scope:** This model's knowledge is based on its foundational English training data. It does not possess advanced or multilingual capabilities.
|
| 53 |
+
* **Potential for Hallucinations:** Like all LLMs, XO can generate incorrect information. **Always verify critical information with a human expert.**
|
| 54 |
+
* **Bias Warning:** The model may reflect biases from its training data.
|
| 55 |
+
|
| 56 |
+
## Citation
|
| 57 |
+
|
| 58 |
+
If you use this model in your research or project, please cite our work:
|
| 59 |
+
|
| 60 |
+
```bibtex
|
| 61 |
+
@misc{saber_xo_3b_2024,
|
| 62 |
+
author = {Saber},
|
| 63 |
+
title = {XO: A Llama 3.2 3B, Unsloth-Trained Cybersecurity Expert},
|
| 64 |
+
year = {2025},
|
| 65 |
+
publisher = {Hugging Face},
|
| 66 |
+
journal = {Hugging Face repository},
|
| 67 |
+
howpublished = {\url{https://huggingface.co/saberbx/XO}}
|
| 68 |
+
}
|