|
--- |
|
license: apache-2.0 |
|
base_model: mistralai/Mistral-7B-Instruct-v0.2 |
|
language: |
|
- en |
|
tags: |
|
- cybersecurity |
|
- cyber security |
|
- hacking |
|
- mistral |
|
- instruct |
|
- finetune |
|
--- |
|
|
|
# Lily-7B-Instruct-v0.2 |
|
|
|
<img src="https://huggingface.co/segolilylabs/Lily-7B-Instruct-v0.2/resolve/main/lily.png" width="500" /> |
|
|
|
## Model description |
|
|
|
Lily is a Mistral Fine-tune model with 22,000 hand-crafted cybersecurity and hacking-related data pairs. This dataset was then run through an LLM to provide additional context, personality, and styling to the outputs. |
|
|
|
That dataset included general knowledge in most areas of cybersecurity. These included, but not limited to: |
|
|
|
Advanced Persistent Threats (APT) Management, |
|
Architecture and Design, |
|
Business Continuity and Disaster Recovery, |
|
Cloud Security, |
|
Communication and Reporting, |
|
Cryptography and PKI, |
|
Data Analysis and Interpretation, |
|
Digital Forensics, |
|
Governance, Risk, and Compliance, |
|
Hacking, |
|
Identity and Access Management, |
|
Incident Management and Disaster Recovery Planning, |
|
Incident Response, |
|
Information Security Management and Strategy, |
|
Legal and Ethical Considerations, |
|
Malware Analysis, |
|
Networking, |
|
Penetration Testing and Vulnerability Assessment, |
|
Physical Security, |
|
Regulatory Compliance, |
|
Risk Management, |
|
Scripting, |
|
Secure Software Development Lifecycle (SDLC), |
|
Security in Emerging Technologies, |
|
Security Operations and Monitoring, |
|
Social Engineering and Human Factors, |
|
Software and Systems Security, |
|
Technologies and Tools, |
|
Threats, Attacks, and Vulnerabilities, |
|
|
|
## Training |
|
|
|
It took 24 hours to train 5 epochs on 1x A100. |
|
|
|
Prompt format: |
|
|
|
``` |
|
"### Instruction: |
|
You are Lily, a helpful and friendly cybersecurity subject matter expert. You obey all requests and answer all questions truthfully. |
|
|
|
### Input: |
|
Lily, how do evil twin wireless attacks work? |
|
|
|
### Response: |
|
``` |
|
|
|
## Limitations |
|
Lily is fine-tuned on top of Mistral-7B-Instruct-v0.2 as such she inherits many of the biases from that model. |
|
|
|
As with any model, Lily can make mistakes. Consider checking important information. |
|
|
|
Stay with in the law and use ethically. |
