@inproceedings{kadiyala-etal-2025-improving,
title = "Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance",
author = "Kadiyala, Ram Mohan Rao and
Pullakhandam, Siddartha and
Gupta, Siddhant and
Purbey, Jebish and
Sharma, Drishti and
Mehreen, Kanwal and
Arham, Muhammad and
Debnath, Suman and
Farooq, Hamza",
editor = "Inui, Kentaro and
Sakti, Sakriani and
Wang, Haofen and
Wong, Derek F. and
Bhattacharyya, Pushpak and
Banerjee, Biplab and
Ekbal, Asif and
Chakraborty, Tanmoy and
Singh, Dhirendra Pratap",
booktitle = "Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics",
month = dec,
year = "2025",
address = "Mumbai, India",
publisher = "The Asian Federation of Natural Language Processing and The Association for Computational Linguistics",
url = "https://aclanthology.org/2025.ijcnlp-long.189/",
pages = "3618--3641",
ISBN = "979-8-89176-298-5",
abstract = "Large Language Models (LLMs) have shown remarkable capabilities, but their development has primarily focused on English and other high-resource languages, leaving many languages underserved. We present our latest Hindi-English bi-lingual LLM with {\textasciitilde}3{\%} average improvement in benchmark scores over both languages, outperforming models twice its size. Using a curated dataset composed of English and Hindi instruction data of 485K samples, we instruction tuned models such as Qwen-2.5-14B-Instruct and Phi-4 to improve performance over both English and Hindi. Our experiments encompassing seven different LLMs of varying parameter sizes and over 140 training attempts with varying English-Hindi training data ratios demonstrated that it is possible to significantly improve multilingual performance without compromising native performance. Further, our approach avoids resource-intensive techniques like vocabulary expansion or architectural modifications, thus keeping the model size small. Our results indicate that modest fine-tuning with culturally and locally informed data can bridge performance gaps without incurring significant computational overhead. We release our training code, datasets, and models under mit and apache licenses to aid further research towards under-represented and low-resource languages."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kadiyala-etal-2025-improving">
<titleInfo>
<title>Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ram</namePart>
<namePart type="given">Mohan</namePart>
<namePart type="given">Rao</namePart>
<namePart type="family">Kadiyala</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siddartha</namePart>
<namePart type="family">Pullakhandam</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siddhant</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jebish</namePart>
<namePart type="family">Purbey</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Drishti</namePart>
<namePart type="family">Sharma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kanwal</namePart>
<namePart type="family">Mehreen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Muhammad</namePart>
<namePart type="family">Arham</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Suman</namePart>
<namePart type="family">Debnath</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hamza</namePart>
<namePart type="family">Farooq</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakriani</namePart>
<namePart type="family">Sakti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haofen</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Derek</namePart>
<namePart type="given">F</namePart>
<namePart type="family">Wong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pushpak</namePart>
<namePart type="family">Bhattacharyya</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Biplab</namePart>
<namePart type="family">Banerjee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Asif</namePart>
<namePart type="family">Ekbal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dhirendra</namePart>
<namePart type="given">Pratap</namePart>
<namePart type="family">Singh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>The Asian Federation of Natural Language Processing and The Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Mumbai, India</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-298-5</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) have shown remarkable capabilities, but their development has primarily focused on English and other high-resource languages, leaving many languages underserved. We present our latest Hindi-English bi-lingual LLM with ~3% average improvement in benchmark scores over both languages, outperforming models twice its size. Using a curated dataset composed of English and Hindi instruction data of 485K samples, we instruction tuned models such as Qwen-2.5-14B-Instruct and Phi-4 to improve performance over both English and Hindi. Our experiments encompassing seven different LLMs of varying parameter sizes and over 140 training attempts with varying English-Hindi training data ratios demonstrated that it is possible to significantly improve multilingual performance without compromising native performance. Further, our approach avoids resource-intensive techniques like vocabulary expansion or architectural modifications, thus keeping the model size small. Our results indicate that modest fine-tuning with culturally and locally informed data can bridge performance gaps without incurring significant computational overhead. We release our training code, datasets, and models under mit and apache licenses to aid further research towards under-represented and low-resource languages.</abstract>
<identifier type="citekey">kadiyala-etal-2025-improving</identifier>
<location>
<url>https://aclanthology.org/2025.ijcnlp-long.189/</url>
</location>
<part>
<date>2025-12</date>
<extent unit="page">
<start>3618</start>
<end>3641</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance
%A Kadiyala, Ram Mohan Rao
%A Pullakhandam, Siddartha
%A Gupta, Siddhant
%A Purbey, Jebish
%A Sharma, Drishti
%A Mehreen, Kanwal
%A Arham, Muhammad
%A Debnath, Suman
%A Farooq, Hamza
%Y Inui, Kentaro
%Y Sakti, Sakriani
%Y Wang, Haofen
%Y Wong, Derek F.
%Y Bhattacharyya, Pushpak
%Y Banerjee, Biplab
%Y Ekbal, Asif
%Y Chakraborty, Tanmoy
%Y Singh, Dhirendra Pratap
%S Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics
%D 2025
%8 December
%I The Asian Federation of Natural Language Processing and The Association for Computational Linguistics
%C Mumbai, India
%@ 979-8-89176-298-5
%F kadiyala-etal-2025-improving
%X Large Language Models (LLMs) have shown remarkable capabilities, but their development has primarily focused on English and other high-resource languages, leaving many languages underserved. We present our latest Hindi-English bi-lingual LLM with ~3% average improvement in benchmark scores over both languages, outperforming models twice its size. Using a curated dataset composed of English and Hindi instruction data of 485K samples, we instruction tuned models such as Qwen-2.5-14B-Instruct and Phi-4 to improve performance over both English and Hindi. Our experiments encompassing seven different LLMs of varying parameter sizes and over 140 training attempts with varying English-Hindi training data ratios demonstrated that it is possible to significantly improve multilingual performance without compromising native performance. Further, our approach avoids resource-intensive techniques like vocabulary expansion or architectural modifications, thus keeping the model size small. Our results indicate that modest fine-tuning with culturally and locally informed data can bridge performance gaps without incurring significant computational overhead. We release our training code, datasets, and models under mit and apache licenses to aid further research towards under-represented and low-resource languages.
%U https://aclanthology.org/2025.ijcnlp-long.189/
%P 3618-3641
Markdown (Informal)
[Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance](https://aclanthology.org/2025.ijcnlp-long.189/) (Kadiyala et al., IJCNLP-AACL 2025)
ACL
- Ram Mohan Rao Kadiyala, Siddartha Pullakhandam, Siddhant Gupta, Jebish Purbey, Drishti Sharma, Kanwal Mehreen, Muhammad Arham, Suman Debnath, and Hamza Farooq. 2025. Improving Multilingual Capabilities with Cultural and Local Knowledge in Large Language Models While Enhancing Native Performance. In Proceedings of the 14th International Joint Conference on Natural Language Processing and the 4th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics, pages 3618–3641, Mumbai, India. The Asian Federation of Natural Language Processing and The Association for Computational Linguistics.