Paper Image

Assessing translation of English and Indian languages with large language models

Published on:

15 November 2023

Primary Category:

Computation and Language

Paper Authors:

Vandan Mujadia,

Ashok Urlana,

Yash Bhaskar,

Penumalla Aditya Pavani,

Kukkapalli Shravya,

Parameswari Krishnamurthy,

Dipti Misra Sharma

Bullets

Key Details

Raw LLMs like LLaMA-2 perform poorly on English-Indian language translation

In-context learning improves performance, but still inadequate for translation

Fine-tuning significantly boosts translation capabilities of LLMs

A 2-stage fine-tuning approach works best for English-to-Indian translation

LLMs show promise for low-resource language translation with minimal data

AI generated summary

Assessing translation of English and Indian languages with large language models

This paper explores using large language models for machine translation between English and 22 Indian languages. The authors evaluate raw LLMs, in-context learning, and fine-tuning approaches. They find that fine-tuned LLaMA models achieve the best performance, producing reasonable translations even for low-resource Indian languages. Their results highlight the potential of LLMs for translation involving underrepresented languages.

Answers from this paper

Comments

No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up