BERT Base for Tigrinya Language

We pre-train a BERT base-uncased model for Tigrinya on a dataset of 40 million tokens trained for 40 epochs.

This repo contains the original pre-trained Flax model that was trained on a TPU v3.8 and its corresponding PyTorch version.

Hyperparameters

The hyperparameters corresponding to the model sizes mentioned above are as follows:

Model Size	L	AH	HS	FFN	P	Seq
BASE	12	12	768	3072	110M	512

(L = number of layers; AH = number of attention heads; HS = hidden size; FFN = feedforward network dimension; P = number of parameters; Seq = maximum sequence length.)

Citation

If you use this model in your product or research, please cite as follows:

@article{Fitsum2021TiPLMs,
  author={Fitsum Gaim and Wonsuk Yang and Jong C. Park},
  title={Monolingual Pre-trained Language Models for Tigrinya},
  year=2021,
  publisher={WiNLP 2021 at EMNLP 2021}
}