File size: 3,141 Bytes
f3e6862
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
87cee01
f3e6862
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
454de0c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
---
language: jv
tags:
- javanese-bert-small-imdb
license: mit
datasets:
- w11wo/imdb-javanese
widget:
- text: "Fast and Furious iku film sing [MASK]."
---

## Javanese BERT Small IMDB
Javanese BERT Small IMDB is a masked language model based on the [BERT model](https://arxiv.org/abs/1810.04805). It was trained on Javanese IMDB movie reviews.

The model was originally the pretrained [Javanese BERT Small model](https://huggingface.co/w11wo/javanese-bert-small) and is later fine-tuned on the Javanese IMDB movie review dataset. It achieved a perplexity of 19.87 on the validation dataset. Many of the techniques used are based on a Hugging Face tutorial [notebook](https://github.com/huggingface/notebooks/blob/master/examples/language_modeling.ipynb) written by [Sylvain Gugger](https://github.com/sgugger).

Hugging Face's `Trainer` class from the [Transformers](https://huggingface.co/transformers) library was used to train the model. PyTorch was used as the backend framework during training, but the model remains compatible with TensorFlow nonetheless.

## Model
| Model                      | #params  | Arch.          | Training/Validation data (text) |
|----------------------------|----------|----------------|---------------------------------|
| `javanese-bert-small-imdb` |   110M   |   BERT Small   | Javanese IMDB (47.5 MB of text) |

## Evaluation Results
The model was trained for 5 epochs and the following is the final result once the training ended.

| train loss | valid loss | perplexity | total time  |
|------------|------------|------------|-------------|
|    3.070   |    2.989   |   19.87    |   3:12:33   |

## How to Use
### As Masked Language Model
```python
from transformers import pipeline

pretrained_name = "w11wo/javanese-bert-small-imdb"

fill_mask = pipeline(
    "fill-mask",
    model=pretrained_name,
    tokenizer=pretrained_name
)

fill_mask("Aku mangan sate ing [MASK] bareng konco-konco")
```
### Feature Extraction in PyTorch
```python
from transformers import BertModel, BertTokenizerFast

pretrained_name = "w11wo/javanese-bert-small-imdb"
model = BertModel.from_pretrained(pretrained_name)
tokenizer = BertTokenizerFast.from_pretrained(pretrained_name)

prompt = "Indonesia minangka negara gedhe."
encoded_input = tokenizer(prompt, return_tensors='pt')
output = model(**encoded_input)
```

## Disclaimer
Do consider the biases which came from the IMDB review that may be carried over into the results of this model.

## Author
Javanese BERT Small was trained and evaluated by [Wilson Wongso](https://w11wo.github.io/). All computation and development are done on Google Colaboratory using their free GPU access.

## Citation

If you use any of our models in your research, please cite:

```bib
@inproceedings{wongso2021causal,
    title={Causal and Masked Language Modeling of Javanese Language using Transformer-based Architectures},
    author={Wongso, Wilson and Setiawan, David Samuel and Suhartono, Derwin},
    booktitle={2021 International Conference on Advanced Computer Science and Information Systems (ICACSIS)},
    pages={1--7},
    year={2021},
    organization={IEEE}
}
```