gpt2-tl-cased

Model Description

As part of the ITANONG project's 10 billion-token Tagalog dataset, we have introduced our initial pre-trained language models for Philippine languages. Our model suite encompasses various BERT-based, GPT-based, and Sentence Transformers tailored for Tagalog,Taglish and Cebuano.

Training Details

This model was trained using an Nvidia V100-32GB GPU on DOST-ASTI Computing and Archiving Research Environment (COARE) - https://asti.dost.gov.ph/projects/coare/

Training Data

The training dataset was compiled from both formal and informal sources, consisting of 5,159,917 instances from formal channels and 3,057,180 from informal sources. More information on pre-processing and training parameters on our paper

Citation

Paper : iTANONG-DS : A Collection of Benchmark Datasets for Downstream Natural Language Processing Tasks on Select Philippine Language

Bibtex:

@inproceedings{visperas-etal-2023-itanong,
    title = "i{TANONG}-{DS} : A Collection of Benchmark Datasets for Downstream Natural Language Processing Tasks on Select {P}hilippine Languages",
    author = "Visperas, Moses L.  and
      Borjal, Christalline Joie  and
      Adoptante, Aunhel John M  and
      Abacial, Danielle Shine R.  and
      Decano, Ma. Miciella  and
      Peramo, Elmer C",
    editor = "Abbas, Mourad  and
      Freihat, Abed Alhakim",
    booktitle = "Proceedings of the 6th International Conference on Natural Language and Speech Processing (ICNLSP 2023)",
    month = dec,
    year = "2023",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.icnlsp-1.34",
    pages = "316--323",
}