Update README.md
Browse files
README.md
CHANGED
|
@@ -1,13 +1,69 @@
|
|
| 1 |
---
|
| 2 |
title: Code Vulnerability Detector
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
-
sdk_version: 6.16.0
|
| 8 |
-
python_version: '3.13'
|
| 9 |
app_file: app.py
|
| 10 |
pinned: false
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: Code Vulnerability Detector
|
| 3 |
+
emoji: π
|
| 4 |
+
colorFrom: red
|
| 5 |
+
colorTo: pink
|
| 6 |
sdk: gradio
|
|
|
|
|
|
|
| 7 |
app_file: app.py
|
| 8 |
pinned: false
|
| 9 |
---
|
| 10 |
|
| 11 |
+
# π Code Vulnerability Detector
|
| 12 |
+
|
| 13 |
+
Deteksi celah keamanan dalam kode secara otomatis menggunakan AI. Model menganalisis kode dan memberikan laporan terstruktur berisi jenis vulnerability, tingkat bahaya, penjelasan, dan saran perbaikan.
|
| 14 |
+
|
| 15 |
+
## π Cara Pakai
|
| 16 |
+
|
| 17 |
+
1. Pilih bahasa pemrograman
|
| 18 |
+
2. Paste kode yang ingin dianalisis
|
| 19 |
+
3. Klik **Analyze Code**
|
| 20 |
+
4. Model akan generate laporan keamanan terstruktur
|
| 21 |
+
|
| 22 |
+
## π Output Format
|
| 23 |
+
|
| 24 |
+
```
|
| 25 |
+
Vulnerability : [tipe vulnerability]
|
| 26 |
+
Severity : [HIGH/MEDIUM/LOW]
|
| 27 |
+
Language : [bahasa pemrograman]
|
| 28 |
+
Penjelasan : [penjelasan kenapa berbahaya]
|
| 29 |
+
Fix : [kode yang sudah diperbaiki]
|
| 30 |
+
```
|
| 31 |
+
|
| 32 |
+
## π οΈ Model Details
|
| 33 |
+
|
| 34 |
+
| Parameter | Value |
|
| 35 |
+
|-----------|-------|
|
| 36 |
+
| Base Model | Qwen/Qwen2.5-Coder-0.5B-Instruct |
|
| 37 |
+
| Method | LoRA fine-tuning |
|
| 38 |
+
| Framework | MLX-LM (Apple Silicon) |
|
| 39 |
+
| Dataset | CyberNative/Code_Vulnerability_Security_DPO |
|
| 40 |
+
| Train Samples | 4,187 |
|
| 41 |
+
| Training Steps | 1,000 |
|
| 42 |
+
| Learning Rate | 2e-4 |
|
| 43 |
+
| Accuracy | 62% |
|
| 44 |
+
|
| 45 |
+
## π Bahasa yang Didukung
|
| 46 |
+
|
| 47 |
+
Python, JavaScript, Java, C++, C#, PHP, Ruby, Swift, Go, Kotlin, Fortran
|
| 48 |
+
|
| 49 |
+
## π Jenis Vulnerability yang Dideteksi
|
| 50 |
+
|
| 51 |
+
- SQL Injection
|
| 52 |
+
- Cross-Site Scripting (XSS)
|
| 53 |
+
- Buffer Overflow
|
| 54 |
+
- Command Injection
|
| 55 |
+
- Improper Authentication
|
| 56 |
+
- Insecure Deserialization
|
| 57 |
+
- Dan banyak lagi...
|
| 58 |
+
|
| 59 |
+
## π Dataset
|
| 60 |
+
|
| 61 |
+
Dataset dari **[CyberNative/Code_Vulnerability_Security_DPO](https://huggingface.co/datasets/CyberNative/Code_Vulnerability_Security_DPO)**:
|
| 62 |
+
- 4,656 pasangan kode vulnerable vs secure
|
| 63 |
+
- 11 bahasa pemrograman
|
| 64 |
+
- Format DPO (Direct Preference Optimization)
|
| 65 |
+
|
| 66 |
+
## π€ Author
|
| 67 |
+
|
| 68 |
+
**Bernadetta**
|
| 69 |
+
[HuggingFace](https://huggingface.co/Bernadetta14)
|