File size: 1,720 Bytes
742319b
 
 
 
 
 
f490665
742319b
d29c8d4
f490665
742319b
2667b61
 
048a511
2667b61
 
 
048a511
6e021b4
481d3f4
 
1a6a62a
481d3f4
d29c8d4
481d3f4
 
1a6a62a
 
 
 
 
 
 
 
481d3f4
6e021b4
 
 
 
 
 
 
 
048a511
6e021b4
9af7991
d85da15
 
 
09f28e7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
---
language: ar
tags: Fill-Mask 
datasets: OSCAR
widget:
- text: " السلام عليكم ورحمة[MASK] وبركاتة"
- text: " اهلا وسهلا بكم في [MASK] من سيربح المليون "
---

# Arabic BERT Model

**AraBERTMo** is an Arabic pre-trained language model based on [Google's BERT architechture](https://github.com/google-research/bert). 
- AraBERTMo_base uses the same BERT-Base config. 

- AraBERTMo_base now comes in 10 new variants
- All models are available on the `HuggingFace` model page under the [Ebtihal](https://huggingface.co/Ebtihal/) name. 
- Checkpoints are available in PyTorch formats.


## Pretraining Corpus

`AraBertMo_base_V1' model was pre-trained on ~3 million words:

- [OSCAR](https://traces1.inria.fr/oscar/) - Arabic version "unshuffled_deduplicated_ar". 


## Training results

this model achieves the following results:

| Task | Num examples | Num Epochs  | Batch Size | steps | Wall time  | training loss| 
|:----:|:----:|:----:|:----:|:-----:|:----:|:-----:|
| Fill-Mask| 10010|  1  | 64 | 157  | 2m 2s | 9.0183  | 


## Load Pretrained Model

You can use this model by installing `torch` or `tensorflow` and Huggingface library `transformers`. And you can use it directly by initializing it like this:  

```python
from transformers import AutoTokenizer, AutoModel
tokenizer = AutoTokenizer.from_pretrained("Ebtihal/AraBertMo_base_V1")
model = AutoModelForMaskedLM.from_pretrained("Ebtihal/AraBertMo_base_V1")
```

 ## This model was built for master's degree research in an organization:
- [University of kufa](https://uokufa.edu.iq/).
- [Faculty of Computer Science and Mathematics](https://mathcomp.uokufa.edu.iq/).
- **Department of Computer Science**