|
|
--- |
|
|
base_model: |
|
|
- mistralai/Mistral-7B-Instruct-v0.2 |
|
|
- segolilylabs/Lily-Cybersecurity-7B-v0.2 |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- cybersecurity |
|
|
- cyber security |
|
|
- hacking |
|
|
- mistral |
|
|
- finetune |
|
|
- ctf |
|
|
datasets: |
|
|
- ShadowEternity/CTF-writeups |
|
|
--- |
|
|
|
|
|
# CTF-Helper |
|
|
|
|
|
Done for ML - Mini Project by: |
|
|
- Akash Varunn |
|
|
- Martin Rozariyo |
|
|
- Praneesh R V |
|
|
- Santosh Tharun |
|
|
- Prasanna Kumar |
|
|
## Model description |
|
|
CTFsolver is a fine-tuned Mistral-7B model, originally adapted from Lily-Cybersecurity-7B-v0.2. This model is specifically trained to assist with Capture The Flag (CTF) challenges, providing detailed writeups, exploit generation, and cybersecurity solutions. |
|
|
|
|
|
Built on top of Mistral-7B, it incorporates cybersecurity knowledge from Lily-Cybersecurity-7B-v0.2 and has been further fine-tuned on real-world CTF challenge solutions and security research papers. |
|
|
|
|
|
|
|
|
## Training |
|
|
|
|
|
CTFsolver has been fine-tuned on ShadowEternity's CTF-writeups dataset, containing: |
|
|
|
|
|
Publicly available CTF solutions |
|
|
|
|
|
Cybersecurity blogs and research papers |
|
|
|
|
|
CTF challenge walkthroughs from top security teams |
|
|
|
|
|
## Limitations |
|
|
Our model is fine-tuned on top of Lily-Cybersecurity which is finetuned from Mistral-7B-Instruct-v0.2 as such it inherits many of the biases from that model. |
|
|
|
|
|
As with any model, It can make mistakes. Consider checking important information. |
|
|
|
|
|
Stay within the law and use ethically. |
