NeoDictaBERT: Pushing the Frontier of BERT models for Hebrew

TL;DR

NeoDictaBERT introduces advanced BERT-style models tailored for Hebrew, outperforming existing models on Hebrew benchmarks and supporting diverse NLP tasks, thus advancing Hebrew language processing research.

Contribution

The paper presents NeoDictaBERT models, adapting modern transformer architectures for Hebrew, and demonstrates their superior performance on multiple Hebrew NLP benchmarks.

Findings

NeoDictaBERT outperforms existing Hebrew models on most benchmarks.

NeoDictaBERT-bilingual excels in retrieval tasks among multilingual models.

Models are publicly released to support Hebrew NLP research.

Abstract

Since their initial release, BERT models have demonstrated exceptional performance on a variety of tasks, despite their relatively small size (BERT-base has ~100M parameters). Nevertheless, the architectural choices used in these models are outdated compared to newer transformer-based models such as Llama3 and Qwen3. In recent months, several architectures have been proposed to close this gap. ModernBERT and NeoBERT both show strong improvements on English benchmarks and significantly extend the supported context window. Following their successes, we introduce NeoDictaBERT and NeoDictaBERT-bilingual: BERT-style models trained using the same architecture as NeoBERT, with a dedicated focus on Hebrew texts. These models outperform existing ones on almost all Hebrew benchmarks and provide a strong foundation for downstream tasks. Notably, the NeoDictaBERT-bilingual model shows strong results on retrieval tasks, outperforming other multilingual models of similar size. In this paper, we describe the training process and report results across various benchmarks. We release the models to the community as part of our goal to advance research and development in Hebrew NLP.