Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- ar
|
4 |
+
|
5 |
+
datasets:
|
6 |
+
- ArSentD-LEV
|
7 |
+
|
8 |
+
tags:
|
9 |
+
- ArSentD-LEV
|
10 |
+
|
11 |
+
|
12 |
+
widget:
|
13 |
+
- text: "يهدي الله من يشاء"
|
14 |
+
- text: "الاسلوب قذر وقمامه"
|
15 |
+
|
16 |
+
---
|
17 |
+
|
18 |
+
# BERT-AJGT
|
19 |
+
Arabic version bert model fine tuned on ArSentD-LEV dataset
|
20 |
+
|
21 |
+
## Data
|
22 |
+
The model were fine-tuned on ~4000 sentence from twitter multiple dialect and five classes we used 3 out of 5 int the experiment.
|
23 |
+
|
24 |
+
0 0.8211 0.8080 0.8145 125
|
25 |
+
1 0.7174 0.7857 0.7500 84
|
26 |
+
2 0.6867 0.6404 0.6628 89
|
27 |
+
|
28 |
+
## Results
|
29 |
+
| class | precision | recall | f1-score | Support |
|
30 |
+
|----------|-----------|--------|----------|---------|
|
31 |
+
| 0 | 0.8211 | 0.8080 | 0.8145 | 125 |
|
32 |
+
| 1 | 0.7174 | 0.7857 | 0.7500 | 84 |
|
33 |
+
| 2 | 0.6867 | 0.6404 | 0.6628 | 89 |
|
34 |
+
| Accuracy | | | 0.7517 | 298 |
|
35 |
+
|
36 |
+
|
37 |
+
## How to use
|
38 |
+
|
39 |
+
You can use these models by installing `torch` or `tensorflow` and Huggingface library `transformers`. And you can use it directly by initializing it like this:
|
40 |
+
|
41 |
+
```python
|
42 |
+
from transformers import AutoModelForSequenceClassification, AutoTokenizer
|
43 |
+
|
44 |
+
model_name="mofawzy/bert-arsentd-lev"
|
45 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_name,num_labels=3)
|
46 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
47 |
+
|
48 |
+
```
|