HeRo: RoBERTa and Longformer Hebrew Language Models

In this paper, we fill in an existing gap in resources available to the Hebrew NLP community by providing it with the largest so far pre-train dataset HeDC4, a state-of-the-art pre-trained language model HeRo for standard length inputs and an efficient transformer Long<PRE_TAG>HeRo</POST_TAG> for long input sequences. The HeRo model was evaluated on the sentiment analysis, the named entity recognition, and the question answering tasks while the Long<PRE_TAG>HeRo</POST_TAG> model was evaluated on the document classification task with a dataset composed of long documents. Both HeRo and Long<PRE_TAG>HeRo</POST_TAG> presented state-of-the-art performance. The dataset and model checkpoints used in this work are publicly available.