Update README.md
Browse files
README.md
CHANGED
@@ -7,8 +7,63 @@ tags:
|
|
7 |
- cybersecurity
|
8 |
- cyber security
|
9 |
- hacking
|
|
|
|
|
|
|
10 |
---
|
11 |
|
12 |
# Lily-7B-Instruct-v0.2
|
13 |
|
14 |
-
<img src="https://huggingface.co/segolilylabs/Lily-7B-Instruct-v0.2/resolve/main/lily.png" width="
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
- cybersecurity
|
8 |
- cyber security
|
9 |
- hacking
|
10 |
+
- mistral
|
11 |
+
- instruct
|
12 |
+
- finetune
|
13 |
---
|
14 |
|
15 |
# Lily-7B-Instruct-v0.2
|
16 |
|
17 |
+
<img src="https://huggingface.co/segolilylabs/Lily-7B-Instruct-v0.2/resolve/main/lily.png" width="600" />
|
18 |
+
|
19 |
+
## Model description
|
20 |
+
|
21 |
+
Lily is a Mistral Fine-tune 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.
|
22 |
+
|
23 |
+
That dataset included general knowledge in most areas on cybersecurity. These included, but not limited to:
|
24 |
+
|
25 |
+
Advanced Persistent Threats (APT) Management
|
26 |
+
Architecture and Design
|
27 |
+
Business Continuity and Disaster Recovery
|
28 |
+
Cloud Security
|
29 |
+
Communication and Reporting
|
30 |
+
Cryptography and PKI
|
31 |
+
Data Analysis and Interpretation
|
32 |
+
Digital Forensics
|
33 |
+
Governance, Risk, and Compliance
|
34 |
+
Hacking
|
35 |
+
Identity and Access Management
|
36 |
+
Incident Management and Disaster Recovery Planning
|
37 |
+
Incident Response
|
38 |
+
Information Security Management and Strategy
|
39 |
+
Legal and Ethical Considerations
|
40 |
+
Malware Analysis
|
41 |
+
Networking
|
42 |
+
Penetration Testing and Vulnerability Assessment
|
43 |
+
Physical Security
|
44 |
+
Regulatory Compliance
|
45 |
+
Risk Management
|
46 |
+
Scripting
|
47 |
+
Secure Software Development Lifecycle (SDLC)
|
48 |
+
Security in Emerging Technologies
|
49 |
+
Security Operations and Monitoring
|
50 |
+
Social Engineering and Human Factors
|
51 |
+
Software and Systems Security
|
52 |
+
Technologies and Tools
|
53 |
+
Threats, Attacks, and Vulnerabilities
|
54 |
+
|
55 |
+
## Training
|
56 |
+
|
57 |
+
It took 24 hours to train 5 epochs on 1x A100.
|
58 |
+
|
59 |
+
Prompt format:
|
60 |
+
|
61 |
+
```
|
62 |
+
"### Instruction:
|
63 |
+
You are Lily, a helpful and friendly cybersecurity subject matter expert. You obey all requests and answer all questions truthfully.
|
64 |
+
|
65 |
+
### Input:
|
66 |
+
Lily, how do evil twin wireless attacks work?
|
67 |
+
|
68 |
+
### Response:
|
69 |
+
```
|