---
language: jv
tags:
- javanese-gpt2-small-imdb-classifier
license: mit
datasets:
- w11wo/imdb-javanese
widget:
- text: "Film sing apik banget!"
---

## Javanese GPT-2 Small IMDB Classifier
Javanese GPT-2 Small IMDB Classifier is a movie-classification model based on the [GPT-2 model](https://cdn.openai.com/better-language-models/language_models_are_unsupervised_multitask_learners.pdf). It was trained on Javanese IMDB movie reviews.

The model was originally [`w11wo/javanese-gpt2-small-imdb`](https://huggingface.co./w11wo/javanese-gpt2-small-imdb) which is then fine-tuned on the [`w11wo/imdb-javanese`](https://huggingface.co./datasets/w11wo/imdb-javanese) dataset consisting of Javanese IMDB movie reviews. It achieved an accuracy of 76.70% 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/text_classification.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-gpt2-small-imdb-classifier` |   124M   |   GPT-2 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 |  accuracy  | total time  |
|------------|------------|------------|-------------|
|    0.324   |    0.574   |   0.767    |   2:0:14    |

## How to Use
### As Text Classifier
```python
from transformers import pipeline

pretrained_name = "w11wo/javanese-gpt2-small-imdb-classifier"

nlp = pipeline(
    "sentiment-analysis",
    model=pretrained_name,
    tokenizer=pretrained_name
)

nlp("Film sing apik banget!")
```

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

## Author
Javanese GPT-2 Small IMDB Classifier 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.