language: jv
tags:
- javanese-gpt2-small-imdb
license: mit
datasets:
- w11wo/imdb-javanese
widget:
- text: Train to Busan yaiku film sing digawe ing Korea Selatan
Javanese GPT-2 Small IMDB
Javanese GPT-2 Small IMDB is a causal language model based on the GPT-2 model. It was trained on Javanese IMDB movie reviews.
The model was originally the pretrained Javanese GPT-2 Small model and is later fine-tuned on the Javanese IMDB movie review dataset. It achieved a perplexity of 55.09 on the validation dataset. Many of the techniques used are based on a Hugging Face tutorial notebook written by Sylvain Gugger.
Hugging Face's Trainer class from the 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 |
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 | perplexity | total time |
|---|---|---|---|
| 3.789 | 4.008 | 55.09 | 1:9:57 |
How to Use (PyTorch)
As Causal Language Model
from transformers import pipeline
pretrained_name = "w11wo/javanese-gpt2-small-imdb"
nlp = pipeline(
"text-generation",
model=pretrained_name,
tokenizer=pretrained_name
)
nlp("Jenengku Budi, saka Indonesia")
Feature Extraction in PyTorch
from transformers import GPT2LMHeadModel, GPT2TokenizerFast
pretrained_name = "w11wo/javanese-gpt2-small-imdb"
model = GPT2LMHeadModel.from_pretrained(pretrained_name)
tokenizer = GPT2TokenizerFast.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 GPT-2 Small was trained and evaluated by Wilson Wongso. All computation and development are done on Google Colaboratory using their free GPU access.