Jais and Jais-chat: Arabic-Centric Foundation and Instruction-Tuned Open Generative Large Language Models

TL;DR

Jais and Jais-chat are Arabic-centric large language models with 13 billion parameters, outperforming existing models in Arabic and showing competitive English performance, aimed at advancing Arabic NLP research.

Contribution

Introduction of the first open Arabic-centric foundation and instruction-tuned LLMs based on GPT-3 architecture, with extensive evaluation and open release for research.

Findings

Superior Arabic language understanding and reasoning

Competitive English performance despite less English data

Open release to foster Arabic NLP research

Abstract

We introduce Jais and Jais-chat, new state-of-the-art Arabic-centric foundation and instruction-tuned open generative large language models (LLMs). The models are based on the GPT-3 decoder-only architecture and are pretrained on a mixture of Arabic and English texts, including source code in various programming languages. With 13 billion parameters, they demonstrate better knowledge and reasoning capabilities in Arabic than any existing open Arabic and multilingual models by a sizable margin, based on extensive evaluation. Moreover, the models are competitive in English compared to English-centric open models of similar size, despite being trained on much less English data. We provide a detailed description of the training, the tuning, the safety alignment, and the evaluation of the models. We release two open versions of the model -- the foundation Jais model, and an instruction-tuned Jais-chat variant -- with the aim of promoting research on Arabic LLMs. Available at https://huggingface.co/inception-mbzuai/jais-13b-chat