Alignment at Pre-training! Towards Native Alignment for Arabic LLMs

TL;DR

This paper introduces the concept of native alignment during pre-training for Arabic LLMs, aiming to improve alignment effectiveness from the outset and enhance model performance and safety.

Contribution

It pioneers the exploration of native alignment in pre-training for Arabic LLMs, demonstrating its benefits through experiments and releasing open-source models.

Findings

Native alignment improves model alignment stability.

Pre-trained Arabic LLMs with native alignment outperform existing models.

Open-source Arabic LLMs achieve state-of-the-art benchmark results.

Abstract

The alignment of large language models (LLMs) is critical for developing effective and safe language models. Traditional approaches focus on aligning models during the instruction tuning or reinforcement learning stages, referred to in this paper as `post alignment'. We argue that alignment during the pre-training phase, which we term `native alignment', warrants investigation. Native alignment aims to prevent unaligned content from the beginning, rather than relying on post-hoc processing. This approach leverages extensively aligned pre-training data to enhance the effectiveness and usability of pre-trained models. Our study specifically explores the application of native alignment in the context of Arabic LLMs. We conduct comprehensive experiments and ablation studies to evaluate the impact of native alignment on model performance and alignment stability. Additionally, we release open-source Arabic LLMs that demonstrate state-of-the-art performance on various benchmarks, providing significant benefits to the Arabic LLM community.