Upload README.md with huggingface_hub
Browse files
README.md
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
---
|
3 |
+
language: en
|
4 |
+
tags:
|
5 |
+
- spam-detection
|
6 |
+
- text-classification
|
7 |
+
- gpt2
|
8 |
+
license: mit
|
9 |
+
---
|
10 |
+
|
11 |
+
# GPT-2 Spam Classifier
|
12 |
+
|
13 |
+
This model is a fine-tuned version of GPT-2 small for spam detection. It was trained on the SMS Spam Collection Dataset.
|
14 |
+
|
15 |
+
## Model Details
|
16 |
+
|
17 |
+
- Base model: GPT-2 small (124M parameters)
|
18 |
+
- Task: Binary classification (spam vs. not spam)
|
19 |
+
- Training Data: SMS Spam Collection Dataset
|
20 |
+
- Fine-tuning approach: Last layer + classification head
|
21 |
+
|
22 |
+
## Usage
|
23 |
+
|
24 |
+
The model expects text input and returns a binary classification (spam/not spam).
|
25 |
+
|
26 |
+
## Performance
|
27 |
+
|
28 |
+
- Training accuracy: ~95%
|
29 |
+
- Validation accuracy: ~95%
|
30 |
+
- Test accuracy: ~93%
|
31 |
+
|
32 |
+
## Limitations
|
33 |
+
|
34 |
+
This model was trained on SMS messages and may not generalize well to other types of text content.
|