Batak Toba language-Indonesian machine translation with transfer learning using No Language Left Behind

This repository contains the implementation of a machine translation model based on the research paper [COMING SOON🤞]. The model architecture and training procedure are described in detail in the provided Jupyter Notebook file.

Requirements

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.13.1
• Tokenizers 0.14.1

Usage

1. Clone this repository to your local machine:

git clone https://github.com/caffeineeee/batak_toba_indonesian_nmt.git
cd batak_toba_indonesian_nmt

2. Open the Jupyter Notebook file batak_toba_indonesian_nmt.ipynb using Jupyter Notebook:

jupyter notebook batak_toba_indonesian_nmt.ipynb

Alternatively, you can run it using Google Colab. For improved training speed, adjust the runtime type to T4 GPU or higher.

3. Follow the instructions in the notebook to train the machine translation model, evaluate its performance, and experiment with different parameters.

License

The code is licensed under the MIT License.