SaulLM-54B和SaulLM-141B：擴展法律領域的領域適應

SaulLM-54B & SaulLM-141B: Scaling Up Domain Adaptation for the Legal Domain

July 28, 2024

作者: Pierre Colombo, Telmo Pires, Malik Boudiaf, Rui Melo, Dominic Culver, Sofia Morgado, Etienne Malaboeuf, Gabriel Hautreux, Johanne Charpentier, Michael Desa

cs.AI

摘要

本文介紹了 SaulLM-54B 和 SaulLM-141B 兩個針對法律領域量身定制的大型語言模型（LLMs）。這些模型分別具有 540 億和 1410 億參數的架構，基於 Mixtral 架構。SaulLM-54B 和 SaulLM-141B 的開發受到大規模領域適應的指導，分為三個策略：（1）利用持續預訓練，包括超過 5400 億法律標記的基本語料庫，（2）實施專門的法律指令遵循協議，以及（3）將模型輸出與法律解釋中的人類偏好對齊。在第二和第三步中整合合成生成的數據增強了模型在解釋和處理法律文本方面的能力，有效地達到了最先進的性能，並在 LegalBench-Instruct 上表現優於先前的開源模型。本研究探討了在這一規模上涉及的特定領域適應中的權衡，提供了可能有助於未來使用強解碼器模型進行領域適應的研究見解。在 SaulLM-7B 的基礎上，本研究改進了方法，以產生一個更適合法律任務的LLM。我們在 SaulLM-54B 和 SaulLM-141B 的基礍上釋出了基本、指令和對齊版本，並採用 MIT 許可證，以促進重複使用和協作研究。

English

In this paper, we introduce SaulLM-54B and SaulLM-141B, two large language models (LLMs) tailored for the legal sector. These models, which feature architectures of 54 billion and 141 billion parameters, respectively, are based on the Mixtral architecture. The development of SaulLM-54B and SaulLM-141B is guided by large-scale domain adaptation, divided into three strategies: (1) the exploitation of continued pretraining involving a base corpus that includes over 540 billion of legal tokens, (2) the implementation of a specialized legal instruction-following protocol, and (3) the alignment of model outputs with human preferences in legal interpretations. The integration of synthetically generated data in the second and third steps enhances the models' capabilities in interpreting and processing legal texts, effectively reaching state-of-the-art performance and outperforming previous open-source models on LegalBench-Instruct. This work explores the trade-offs involved in domain-specific adaptation at this scale, offering insights that may inform future studies on domain adaptation using strong decoder models. Building upon SaulLM-7B, this study refines the approach to produce an LLM better equipped for legal tasks. We are releasing base, instruct, and aligned versions on top of SaulLM-54B and SaulLM-141B under the MIT License to facilitate reuse and collaborative research.