Noam Ordan

Amir Zeldes, Nick Howell, Noam Ordan et al.

Foundational Hebrew NLP tasks such as segmentation, tagging and parsing, have relied to date on various versions of the Hebrew Treebank (HTB, Sima'an et al. 2001). However, the data in HTB, a single-source newswire corpus, is now over 30 years old, and does not cover many aspects of contemporary Hebrew on the web. This paper presents a new, freely available UD treebank of Hebrew stratified from a range of topics selected from Hebrew Wikipedia. In addition to introducing the corpus and evaluating the quality of its annotations, we deploy automatic validation tools based on grew (Guillaume, 2021), and conduct the first cross domain parsing experiments in Hebrew. We obtain new state-of-the-art (SOTA) results on UD NLP tasks, using a combination of the latest language modelling and some incremental improvements to existing transformer based approaches. We also release a new version of the UD HTB matching annotation scheme updates from our new corpus.

Noam Kayzer, Dan Revital, Ori Bar Joseph et al.

We present Hebatron, a Hebrew-specialized open-weight large language model built on the NVIDIA Nemotron-3 sparse Mixture-of-Experts architecture. Training employs a three-phase easy-to-hard curriculum with continuous anti-forgetting anchoring, followed by supervised fine-tuning on 2 million bilingual Hebrew--English samples. The curriculum ordering alone yields a 3-point aggregate benchmark gain over the reversed configuration. Hebatron achieves a Hebrew reasoning average of 73.8\%, outperforming DictaLM-3.0-24B-Thinking (68.9\%) and remaining competitive with Gemma-3-27B-IT on GSM8K-HE and Israeli Trivia, while activating only 3B parameters per forward pass across a 30B-parameter model, delivering approximately 9 times higher inference throughput at native context lengths up to 65,536 tokens. To our knowledge, this is the first language-specific adaptation of the Nemotron-3 architecture for any target language, and the first open-weight Hebrew-specialized MoE model with native long-context support. Model weights are released openly to support further research in Hebrew and Semitic-language NLP.

Ella Rabinovich, Noam Ordan, Shuly Wintner

Translation has played an important role in trade, law, commerce, politics, and literature for thousands of years. Translators have always tried to be invisible; ideal translations should look as if they were written originally in the target language. We show that traces of the source language remain in the translation product to the extent that it is possible to uncover the history of the source language by looking only at the translation. Specifically, we automatically reconstruct phylogenetic language trees from monolingual texts (translated from several source languages). The signal of the source language is so powerful that it is retained even after two phases of translation. This strongly indicates that source language interference is the most dominant characteristic of translated texts, overshadowing the more subtle signals of universal properties of translation.

Ella Rabinovich, Sergiu Nisioi, Noam Ordan et al.

We present a computational analysis of three language varieties: native, advanced non-native, and translation. Our goal is to investigate the similarities and differences between non-native language productions and translations, contrasting both with native language. Using a collection of computational methods we establish three main results: (1) the three types of texts are easily distinguishable; (2) non-native language and translations are closer to each other than each of them is to native language; and (3) some of these characteristics depend on the source or native language, while others do not, reflecting, perhaps, unified principles that similarly affect translations and non-native language.