From Scratch to Fine-Tuned: A Comparative Study of Transformer Training Strategies for Legal Machine Translation

TL;DR

This study compares transformer training strategies for legal machine translation, showing that fine-tuning pre-trained models significantly improves translation quality for English-Hindi legal documents.

Contribution

It provides a comparative analysis of from-scratch training versus fine-tuning pre-trained models for legal machine translation, highlighting the effectiveness of domain adaptation.

Findings

Fine-tuned OPUS-MT achieved a SacreBLEU score of 46.03.

Fine-tuning outperformed models trained from scratch.

Demonstrates potential for improved legal translation in multilingual contexts.

Abstract

In multilingual nations like India, access to legal information is often hindered by language barriers, as much of the legal and judicial documentation remains in English. Legal Machine Translation (L-MT) offers a scalable solution to this challenge by enabling accurate and accessible translations of legal documents. This paper presents our work for the JUST-NLP 2025 Legal MT shared task, focusing on English-Hindi translation using Transformer-based approaches. We experiment with 2 complementary strategies, fine-tuning a pre-trained OPUS-MT model for domain-specific adaptation and training a Transformer model from scratch using the provided legal corpus. Performance is evaluated using standard MT metrics, including SacreBLEU, chrF++, TER, ROUGE, BERTScore, METEOR, and COMET. Our fine-tuned OPUS-MT model achieves a SacreBLEU score of 46.03, significantly outperforming both baseline and from-scratch models. The results highlight the effectiveness of domain adaptation in enhancing translation quality and demonstrate the potential of L-MT systems to improve access to justice and legal transparency in multilingual contexts.