ARBERT is one of two models described in the paper "ARBERT & MARBERT: Deep Bidirectional Transformers for Arabic". ARBERT is a large-scale pre-trained masked language model focused on Modern Standard Arabic (MSA). To train ARBERT, we use the same architecture as BERT-base: 12 attention layers, each has 12 attention heads and 768 hidden dimensions, a vocabulary of 100K WordPieces, making ∼163M parameters. We train ARBERT on a collection of Arabic datasets comprising 61GB of text (6.2B tokens). For more information, please visit our own GitHub repo.