Papers by Alham Fikri Aji

In Neural Machine Translation, What Does Transfer Learning Transfer? (2020.acl-main)

Challenge:	In recent years, neural network models have grown dramatically in terms of number of parameters, so exchanging gradients during data-parallel training is costly in terms both of bandwidth and time.
Approach:	They propose to combine the compressed global gradient with the local gradient to restore Transformer convergence while RNNs converge faster.
Outcome:	The proposed method restores transformer convergence while RNNs converge faster.

Copied to clipboard

Alham Fikri Aji, Nikolay Bogoychev, Kenneth Heafield, Rico Sennrich

Thank You, Stingray: Multilingual Large Language Models Can Not (Yet) Disambiguate Cross-Lingual Word Senses (2025.findings-naacl)

Challenge:	a recent study found that word embeddings are not necessary for transfer learning.
Approach:	They perform several ablation studies that limit information transfer and measure the quality impact across three language pairs to gain a black-box understanding of transfer learning.
Outcome:	The proposed method can eliminate the need for a warm-up phase when training transformer models in high resource language pairs.

Copied to clipboard

Samuel Cahyawijaya, Ruochen Zhang, Jan Christian Blaise Cruz, Holy Lovenia, Elisa Gilbert, Hiroki Nomoto, Alham Fikri Aji

Statement-Tuning Enables Efficient Cross-lingual Generalization in Encoder-only Models (2025.findings-acl)

Challenge:	Existing studies on multilingual large language models have raised concerns about their reliability beyond English.
Approach:	They propose a benchmark for cross-lingual sense disambiguation that uses false friends to identify the limitation of cross-linguistic sense disembarrassment in LLMs.
Outcome:	The proposed benchmark pinpoints the limitation of cross-lingual sense disambiguation in LLMs by using false friends in four languages.

Copied to clipboard

Ahmed Elshabrawy, Thanh-Nhi Nguyen, Yeeun Kang, Lihan Feng, Annant Jain, Faadil Abdullah Shaikh, Jonibek Mansurov, Mohamed Fazli Mohamed Imam, Jesus-German Ortiz-Barajas, Rendi Chevi, Alham Fikri Aji

Responsible & Trustworthy NLP

Challenge:	Large Language Models excel in zero-shot and few-shot tasks, but their architecture makes them difficult to use.
Approach:	They adapt Large Language Models (LLMs) for zero-shot generalization using Statement Tuning . they find encoders can achieve zero- shot cross-lingual generalization .
Outcome:	The proposed model generalizes well across languages while being more efficient.

NusaX: Multilingual Parallel Sentiment Dataset for 10 Indonesian Local Languages (2023.eacl-main)

Copied to clipboard

Genta Indra Winata, Alham Fikri Aji, Samuel Cahyawijaya, Rahmad Mahendra, Fajri Koto, Ade Romadhony, Kemal Kurniawan, David Moeljadi, Radityo Eko Prasojo, Pascale Fung, Timothy Baldwin, Jey Han Lau, Rico Sennrich, Sebastian Ruder

Data Laundering: Artificially Boosting Benchmark Results through Knowledge Distillation (2025.acl-long)

Challenge:	In Indonesia, many languages are endangered and some are even extinct due to the unavailability of data resources and benchmarks.
Approach:	They propose a high-quality multilingual parallel corpus that covers 10 local languages from Indonesia.
Outcome:	The proposed resource includes sentiment and machine translation datasets, and bilingual lexicons.

Copied to clipboard

Jonibek Mansurov, Akhmed Sakip, Alham Fikri Aji

One Country, 700+ Languages: NLP Challenges for Underrepresented Languages and Dialects in Indonesia (2022.acl-long)

Challenge:	Existing studies show that language model benchmarks are vulnerable to manipulation and exploitation.
Approach:	They propose a method that allows the covert transfer of benchmark-specific knowledge through seemingly legitimate intermediate training steps.
Outcome:	The proposed method can achieve significant improvements in accuracy without developing reasoning capabilities.

Copied to clipboard

Alham Fikri Aji, Genta Indra Winata, Fajri Koto, Samuel Cahyawijaya, Ade Romadhony, Rahmad Mahendra, Kemal Kurniawan, David Moeljadi, Radityo Eko Prasojo, Timothy Baldwin, Jey Han Lau, Sebastian Ruder

Responsible & Trustworthy NLP

Challenge:	There are more than 700 languages spoken in Indonesia, equal to 10% of the world's languages, second only to Papua New Guinea.
Approach:	They focus on the languages spoken in Indonesia, the world's second most linguistically diverse nation, and the fourth most populous nation of the world.
Outcome:	The proposed model is based on the languages spoken in Indonesia, the world's second-most linguistically diverse nation, with 273 million people spread over 17,508 islands.

IndoCollex: A Testbed for Morphological Transformation of Indonesian Colloquial Words (2021.findings-acl)

Copied to clipboard

Haryo Akbarianto Wibowo, Made Nindyatama Nityasya, Afra Feyza Akyürek, Suci Fitriany, Alham Fikri Aji, Radityo Eko Prasojo, Derry Tanti Wijaya

Syntactic Text Processing

Challenge:	Existing research on word normalization in Indonesian language relies on static dictionaries and machine translation.
Approach:	They propose to use Twitter to annotate Indonesian colloquial words with their standard forms and their word formation types/tags to perform morphological word normalization.
Outcome:	The proposed dataset analyzes morphological word normalization on Indonesian colloquial Lexicons and provides a baseline for future work.

Enabling Natural Zero-Shot Prompting on Encoder Models via Statement-Tuning (2025.findings-naacl)

Copied to clipboard

Ahmed Elshabrawy, Yongxin Huang, Iryna Gurevych, Alham Fikri Aji

MoMentS: A Comprehensive Multimodal Benchmark for Theory of Mind (2025.findings-emnlp)

Challenge:	Large Language Models (LLMs) exhibit remarkable capabilities in zero-shot and few-shot settings, but they struggle with extending to few- shot and zero- shot settings due to their architectural design.
Approach:	They propose a technique that models discriminative tasks as a set of finite statements and trains an encoder model to discriminate between the potential statements to determine the label.
Outcome:	The proposed method achieves competitive performance compared to state-of-the-art LLMs with significantly fewer parameters.

Copied to clipboard

Emilio Villa-Cueva, S M Masrur Ahmed, Rendi Chevi, Jan Christian Blaise Cruz, Kareem Elzeky, Fermin Cristobal, Alham Fikri Aji, Skyler Wang, Rada Mihalcea, Thamar Solorio

Macaron: Controlled, Human-Written Benchmark for Multilingual and Multicultural Reasoning via Template-Filling (2026.acl-long)

Challenge:	MoMentS is a benchmark designed to assess the ToM capabilities of multimodal large language models (LLMs) in short films.
Approach:	They introduce a benchmark to assess the ToM capabilities of multimodal large language models (LLMs) through realistic, narrative-rich scenarios presented in short films.
Outcome:	The proposed benchmark features long video context windows and realistic social interactions that provide deeper insight into characters’ mental states.

Copied to clipboard

Alaa Elsetohy, Sama Hadhoud, Haryo Akbarianto Wibowo, Chenxi Whitehouse, Genta Indra Winata, Fajri Koto, Alham Fikri Aji

Reasoning

Challenge:	Existing benchmarks test reasoning over culturally grounded premises, but translation-parallel benchmarks inherit English-centric scenarios.
Approach:	They propose a template-first benchmark that factorizes reasoning type and cultural aspect across question languages.
Outcome:	The proposed benchmark factorizes reasoning type and cultural aspect across question languages.

A Multi-Labeled Dataset for Indonesian Discourse: Examining Toxicity, Polarization, and Demographics Information (2025.findings-acl)

Copied to clipboard

Lucky Susanto, Musa Izzanardi Wijanarko, Prasetia Anugrah Pratama, Zilu Tang, Fariz Akyas, Traci Hong, Ika Karlina Idris, Alham Fikri Aji, Derry Tanti Wijaya

Text Classification

Challenge:	Prior research has focused on toxicity and polarization as separate problems . extreme polarizing deepens divisions, often leading to hostility and fragmentation .
Approach:	They propose to use a multi-label Indonesian dataset annotated for toxicity, polarization, and annotator demographic information to study polarizing language and toxicity.
Outcome:	The proposed dataset shows that polarization cues improve toxicity classification and vice versa.

KazMMLU: Evaluating Language Models on Kazakh, Russian, and Regional Knowledge of Kazakhstan (2025.acl-long)

Copied to clipboard

Mukhammed Togmanov, Nurdaulet Mukhituly, Diana Turmakhan, Jonibek Mansurov, Maiya Goloburda, Akhmed Sakip, Zhuohan Xie, Yuxia Wang, Bekassyl Syzdykov, Nurkhan Laiyk, Alham Fikri Aji, Ekaterina Kochmar, Preslav Nakov, Fajri Koto

Afri-MCQA: Multimodal Cultural Question Answering for African Languages (2026.acl-long)

Challenge:	Kazakh language remains underrepresented in the field of natural language processing despite the country's population exceeding twenty million . however, there is a lack of dedicated models and benchmark evaluations specifically tailored to Kazakh languages.
Approach:	They propose to create a dataset specifically designed for Kazakh language with 23,000 questions sourced from authentic educational materials and manually validated by native speakers and educators.
Outcome:	The first MMLU-style dataset specifically designed for Kazakh language.

Copied to clipboard

Atnafu Lambebo Tonja, Srija Anand, Emilio Villa-Cueva, Israel Abebe Azime, Jesujoba Oluwadara Alabi, Muhidin A. Mohamed, Debela Desalegn Yadeta, Negasi Haile Abadi, Abigail Oppong, Nnaemeka Casmir Obiefuna, Idris Abdulmumin, Naome A Etori, Eric Peter Wairagala, Kanda Patrick Tshinu, Imanigirimbabazi Emmanuel, Gabofetswe Malema, Alham Fikri Aji, David Ifeoluwa Adelani, Thamar Solorio

NusaCrowd: Open Source Initiative for Indonesian NLP Resources (2023.findings-acl)

Challenge:	Afri-MCQA is the first multilingual cultural question-answering benchmark containing 7.5k Q A pairs across 15 African languages from 12 countries.
Approach:	They introduce Afri-MCQA, the first multilingual cultural question-answering benchmark containing 7.5k Q A pairs across 15 African languages from 12 countries.
Outcome:	The proposed model shows poor performance across cultures, with near zero accuracy on open-ended VQA when queried through native language or speech.

Copied to clipboard

Samuel Cahyawijaya, Holy Lovenia, Alham Fikri Aji, Genta Winata, Bryan Wilie, Fajri Koto, Rahmad Mahendra, Christian Wibisono, Ade Romadhony, Karissa Vincentio, Jennifer Santoso, David Moeljadi, Cahya Wirawan, Frederikus Hudi, Muhammad Satrio Wicaksono, Ivan Parmonangan, Ika Alfina, Ilham Firdausi Putra, Samsul Rahmadani, Yulianti Oenang, Ali Septiandri, James Jaya, Kaustubh Dhole, Arie Suryani, Rifki Afina Putri, Dan Su, Keith Stevens, Made Nindyatama Nityasya, Muhammad Adilazuarda, Ryan Hadiwijaya, Ryandito Diandaru, Tiezheng Yu, Vito Ghifari, Wenliang Dai, Yan Xu, Dyah Damapuspita, Haryo Wibowo, Cuk Tho, Ichwanul Karo Karo, Tirana Fatyanosa, Ziwei Ji, Graham Neubig, Timothy Baldwin, Sebastian Ruder, Pascale Fung, Herry Sujaini, Sakriani Sakti, Ayu Purwarianti

Challenge:	Existing NLP research in Indonesian languages has been held back by factors such as language diversity, orthographic variation, resource limitation and other societal challenges.
Approach:	They present a collaborative initiative to collect and unify existing resources for Indonesian languages and open access to previously non-public resources.
Outcome:	The results show that the datasets are highly reliable and can be used to generate the first zero-shot benchmarks for natural language understanding and generation in Indonesian and the local languages of Indonesia.

Multi-lingual and Multi-cultural Figurative Language Understanding (2023.findings-acl)

Copied to clipboard

Anubha Kabra, Emmy Liu, Simran Khanuja, Alham Fikri Aji, Genta Winata, Samuel Cahyawijaya, Anuoluwapo Aremu, Perez Ogayo, Graham Neubig

Sentiment Analysis

Challenge:	Figures permeate human communication, but are understudied in NLP.
Approach:	They create a figurative language inference dataset for seven languages associated with a variety of cultures, using cultural and regional concepts for figurativ expressions.
Outcome:	The results show that the most common figurative expressions are found in Hindi, Indonesian, Javanese, Kannada, Sundanese, Swahili and Yoruba.

Is Human-Like Text Liked by Humans? Multilingual Human Detection and Preference Against AI (2026.acl-long)

Copied to clipboard

Yuxia Wang, Rui Xing, Jonibek Mansurov, Giovanni Puccetti, Zhuohan Xie, Minh Ngoc Ta, Jiahui Geng, Jinyan Su, Mervat Abassy, Saadeldine Eletter, Kareem Elozeiri, Nurkhan Laiyk, Maiya Goloburda, Tarek Mahmoud, Raj Vardhan Tomar, Alexander Aziz, Ryuto Koike, Masahiro Kaneko, Artem Shelmanov, Ekaterina Artemova, Vladislav Mikhailov, Akim Tsvigun, Alham Fikri Aji, Nizar Habash, Iryna Gurevych, Preslav Nakov

Direct Fact Retrieval from Knowledge Graphs without Entity Linking (2023.acl-long)

Challenge:	Prior studies have shown that distinguishing text generated by Large Language Models from human-written text is challenging for humans and often no better than random guessing.
Approach:	They conduct extensive case study to determine the upper bound of human detection accuracy.
Outcome:	The findings challenge previous conclusions on human detection accuracy across languages and domains.

Copied to clipboard

Jinheon Baek, Alham Fikri Aji, Jens Lehmann, Sung Ju Hwang

WorldCuisines: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines (2025.naacl-long)

Challenge:	Existing methods to retrieve facts from Knowledge Graphs (KGs) require additional labels and may accumulate errors .
Approach:	They propose a framework that directly retrieves facts from KGs given input text based on their representational similarities.
Outcome:	The proposed framework outperforms baselines on multiple fact retrieval tasks.

Copied to clipboard

Genta Indra Winata, Frederikus Hudi, Patrick Amadeus Irawan, David Anugraha, Rifki Afina Putri, Wang Yutong, Adam Nohejl, Ubaidillah Ariq Prathama, Nedjma Ousidhoum, Afifa Amriani, Anar Rzayev, Anirban Das, Ashmari Pramodya, Aulia Adila, Bryan Wilie, Candy Olivia Mawalim, Cheng Ching Lam, Daud Abolade, Emmanuele Chersoni, Enrico Santus, Fariz Ikhwantri, Garry Kuwanto, Hanyang Zhao, Haryo Akbarianto Wibowo, Holy Lovenia, Jan Christian Blaise Cruz, Jan Wira Gotama Putra, Junho Myung, Lucky Susanto, Maria Angelica Riera Machin, Marina Zhukova, Michael Anugraha, Muhammad Farid Adilazuarda, Natasha Christabelle Santosa, Peerat Limkonchotiwat, Raj Dabre, Rio Alexander Audino, Samuel Cahyawijaya, Shi-Xiong Zhang, Stephanie Yulia Salim, Yi Zhou, Yinxuan Gui, David Ifeoluwa Adelani, En-Shiun Annie Lee, Shogo Okada, Ayu Purwarianti, Alham Fikri Aji, Taro Watanabe, Derry Tanti Wijaya, Alice Oh, Chong-Wah Ngo

Visual Data in NLP

Challenge:	Vision Language Models struggle with cultural-specific knowledge, especially in languages other than English and in underrepresented cultural contexts.
Approach:	They propose a visual question answering (VQA) dataset with text-image pairs across 30 languages and dialects and a training dataset.
Outcome:	The proposed model performs better with correct location context, but struggles with adversarial contexts and predicting specific regional cuisines and languages.

IndoNLI: A Natural Language Inference Dataset for Indonesian (2021.emnlp-main)

Copied to clipboard

Rahmad Mahendra, Alham Fikri Aji, Samuel Louvan, Fahrurrozi Rahman, Clara Vania

Reasoning

Challenge:	XLM-R model outperforms other pre-trained models in annotated data.
Approach:	They adapt the data collection protocol for MNLI and collect 18K sentence pairs annotated by crowd workers and experts.
Outcome:	The proposed dataset outperforms other pre-trained models on the expert-annotated data.

CaMMT: Benchmarking Culturally Aware Multimodal Machine Translation (2025.findings-emnlp)

Copied to clipboard

Emilio Villa-Cueva, Sholpan Bolatzhanova, Diana Turmakhan, Kareem Elzeky, Henok Biadglign Ademtew, Alham Fikri Aji, Vladimir Araujo, Israel Abebe Azime, Jinheon Baek, Frederico Belcavello, Fermin Cristobal, Jan Christian Blaise Cruz, Mary Dabre, Raj Dabre, Toqeer Ehsan, Naome A Etori, Fauzan Farooqui, Jiahui Geng, Guido Ivetta, Thanmay Jayakumar, Soyeong Jeong, Zheng Wei Lim, Aishik Mandal, Sofía Martinelli, Mihail Minkov Mihaylov, Daniil Orel, Aniket Pramanick, Sukannya Purkayastha, Israfel Salazar, Haiyue Song, Tiago Timponi Torrent, Debela Desalegn Yadeta, Injy Hamed, Atnafu Lambebo Tonja, Thamar Solorio

Do Language Models Understand Honorific Systems in Javanese? (2025.acl-long)

Challenge:	a human-curated benchmark of over 5,800 triples of images is used to evaluate multimodal translation systems.
Approach:	They introduce a human-curated benchmark of over 5,800 triples of images along with parallel captions in English and regional languages.
Outcome:	The results show that visual context improves translation quality in culturally-specific items .

Copied to clipboard

Mohammad Rifqi Farhansyah, Iwan Darmawan, Adryan Kusumawardhana, Genta Indra Winata, Alham Fikri Aji, Derry Tanti Wijaya

SEA-Guard: Culturally Grounded Multilingual Safeguard for Southeast Asia (2026.findings-acl)

Challenge:	Despite its cultural and linguistic significance, there has been limited progress in developing a comprehensive corpus to capture these variations for natural language processing (NLP) tasks.
Approach:	They propose to use a dataset to capture the nuances of Unggah-Ungguh Basa, the Javanese speech etiquette framework, to assess the ability of language models to process various levels of Javanesi honorifics.
Outcome:	The proposed dataset encapsulates the nuances of Unggah-Ungguh Basa, the Javanese speech etiquette framework.

Copied to clipboard

Panuthep Tasawong, Jian Gang Ngui, Alham Fikri Aji, Trevor Cohn, Peerat Limkonchotiwat

Making Asynchronous Stochastic Gradient Descent Work for Transformers (D19-56)

Challenge:	Existing safeguard models rely on translation of English datasets, missing regional and cultural nuances.
Approach:	They propose a framework to generate culturally grounded safety datasets for Southeast Asia . SEA-Guard family is the first multilingual safeguard model grounded in SEA cultural contexts .
Outcome:	The proposed model outperforms existing safeguard models in detecting regionally sensitive content while maintaining strong general safety performance.

Copied to clipboard

Alham Fikri Aji, Kenneth Heafield

Accelerating Asynchronous Stochastic Gradient Descent for Neural Machine Translation (D18-1)

Challenge:	Asynchronous stochastic gradient descent (SGD) converges poorly for Transformer models . synchronous SGD is faster at raw training speed since it avoids waiting for synchronization .
Approach:	They propose a method to restore convergence by summing several asynchronous updates instead of applying them immediately.
Outcome:	The proposed method achieves the same BLEU score 1.36 times faster than asynchronous SGD.

Copied to clipboard

Nikolay Bogoychev, Kenneth Heafield, Alham Fikri Aji, Marcin Junczys-Dowmunt

Softpick: No Attention Sink, No Massive Activations with Rectified Softmax (2026.findings-acl)

Challenge:	In order to achieve faster training we increase the mini-batch size and scale the learning rate accordingly.
Approach:	They propose a technique that delays gradient updates by increasing the mini-batch size to improve the model's convergence.
Outcome:	The proposed technique can train a shallow machine translation system 27% faster than an optimized baseline with negligible penalty in BLEU.

Copied to clipboard

Zayd Muhammad Kawakibi Zuhri, Erland Hilman Fuadi, Alham Fikri Aji

Crowdsource, Crawl, or Generate? Creating SEA-VL, a Multicultural Vision-Language Dataset for Southeast Asia (2025.acl-long)

Challenge:	Quantized models using softpick outperform softmax on standard benchmarks . softmax is widely used in statistics and especially in machine learning .
Approach:	They introduce a rectified, not sum-to-one, drop-in replacement for softmax in transformer attention mechanisms that eliminates attention sink and massive activations.
Outcome:	The proposed model outperforms softmax on benchmarks with lower bit precisions.

Copied to clipboard

Samuel Cahyawijaya, Holy Lovenia, Joel Ruben Antony Moniz, Tack Hwa Wong, Mohammad Rifqi Farhansyah, Thant Thiri Maung, Frederikus Hudi, David Anugraha, Muhammad Ravi Shulthan Habibi, Muhammad Reza Qorib, Amit Agarwal, Joseph Marvin Imperial, Hitesh Laxmichand Patel, Vicky Feliren, Bahrul Ilmi Nasution, Manuel Antonio Rufino, Genta Indra Winata, Rian Adam Rajagede, Carlos Rafael Catalan, Mohamed Fazli Mohamed Imam, Priyaranjan Pattnayak, Salsabila Zahirah Pranida, Kevin Pratama, Yeshil Bangera, Adisai Na-Thalang, Patricia Nicole Monderin, Yueqi Song, Christian Simon, Lynnette Hui Xian Ng, Richardy Lobo Sapan, Taki Hasan Rafi, Bin Wang, null Supryadi, Kanyakorn Veerakanjana, Piyalitt Ittichaiwong, Matthew Theodore Roque, Karissa Vincentio, Takdanai Kreangphet, Phakphum Artkaew, Kadek Hendrawan Palgunadi, Yanzhi Yu, Rochana Prih Hastuti, William Nixon, Mithil Bangera, Adrian Xuan Wei Lim, Aye Hninn Khine, Hanif Muhammad Zhafran, Teddy Ferdinan, Audra Aurora Izzani, Ayushman Singh, Evan Evan, Jauza Akbar Krito, Michael Anugraha, Fenal Ashokbhai Ilasariya, Haochen Li, John Amadeo Daniswara, Filbert Aurelian Tjiaranata, Eryawan Presma Yulianrifat, Can Udomcharoenchaikit, Fadil Risdian Ansori, Mahardika Krisna Ihsani, Giang Nguyen, Anab Maulana Barik, Dan John Velasco, Rifo Ahmad Genadi, Saptarshi Saha, Chengwei Wei, Isaiah Edri W. Flores, Kenneth Chen Ko Han, Anjela Gail D. Santos, Wan Shen Lim, Kaung Si Phyo, Tim Santos, Meisyarah Dwiastuti, Jiayun Luo, Jan Christian Blaise Cruz, Ming Shan Hee, Ikhlasul Akmal Hanif, M.Alif Al Hakim, Muhammad Rizky Sya’ban, Kun Kerdthaisong, Lester James Validad Miranda, Fajri Koto, Tirana Noor Fatyanosa, Alham Fikri Aji, Jostin Jerico Rosal, Jun Kevin, Robert Wijaya, Onno P. Kampman, Ruochen Zhang, Börje F. Karlsson, Peerat Limkonchotiwat

Crosslingual Generalization through Multitask Finetuning (2023.acl-long)

Challenge:	Southeast Asia is underrepresented in vision-language research . SEA-VL is an open-source initiative dedicated to developing culturally relevant datasets for SEA languages.
Approach:	They propose to use crowdsourced, automated image crawling and synthetic image generation to develop culturally relevant datasets for SEA languages.
Outcome:	The proposed datasets capture SEA cultural nuances and contexts better than existing datasets.

Copied to clipboard

Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng Xin Yong, Hailey Schoelkopf, Xiangru Tang, Dragomir Radev, Alham Fikri Aji, Khalid Almubarak, Samuel Albanie, Zaid Alyafeai, Albert Webson, Edward Raff, Colin Raffel

WangchanThaiInstruct: An instruction-following Dataset for Culture-Aware, Multitask, and Multi-domain Evaluation in Thai (2025.emnlp-main)

Challenge:	Multitask prompted finetuning (MTF) has been shown to help large language models generalize to new tasks in a zero-shot setting, but so far explorations of MTF have focused on English data and models.
Approach:	They apply multitask prompted finetuning to pretrained multilingual models and generate variants called BLOOMZ and mT0.
Outcome:	The proposed models can generalize to non-English languages that have never been seen before.

Copied to clipboard

Peerat Limkonchotiwat, Pume Tuchinda, Lalita Lowphansirikul, Surapon Nonesung, Panuthep Tasawong, Alham Fikri Aji, Can Udomcharoenchaikit, Sarana Nutanong

Machine Translation

Challenge:	Existing benchmarks for large language models rely on translations, missing cultural and domain specificity.
Approach:	They present a human-authored dataset for evaluation and instruction tuning in Thai . findings highlight need for culturally and professionally grounded instruction data .
Outcome:	a human-authored dataset for evaluation and instruction tuning in Thai outperforms translation-based models . findings highlight need for culturally and professionally grounded instruction data .

Marian: Fast Neural Machine Translation in C++ (P18-4)

Copied to clipboard

Marcin Junczys-Dowmunt, Roman Grundkiewicz, Tomasz Dwojak, Hieu Hoang, Kenneth Heafield, Tom Neckermann, Frank Seide, Ulrich Germann, Alham Fikri Aji, Nikolay Bogoychev, André F. T. Martins, Alexandra Birch

Machine Translation

Challenge:	In this paper, we present Marian, an efficient and self-contained Neural Machine Translation framework . Marian is written in pure C++ with minimal dependencies .
Approach:	They present Marian, an efficient and self-contained Neural Machine Translation framework written in pure C++ with minimal dependencies.
Outcome:	The proposed framework achieves high training and translation speed with minimal dependencies . it is currently being deployed in multiple European projects .

From Surveys to Narratives: Rethinking Cultural Value Adaptation in LLMs (2025.emnlp-main)

Copied to clipboard

Farid Adilazuarda, Chen Cecilia Liu, Iryna Gurevych, Alham Fikri Aji

Text Classification

Challenge:	Adapting cultural values in Large Language Models presents significant challenges due to biases and data limitations.
Approach:	They propose to augment World Values Survey (WVS) data with encyclopedic and scenario-based cultural narratives from Wikipedia and NormAd to address these limitations.
Outcome:	The proposed approach enhances cultural distinctiveness and improves classification performance across cultures.

NusaAksara: A Multimodal and Multilingual Benchmark for Preserving Indonesian Indigenous Scripts (2025.acl-long)

Copied to clipboard

Muhammad Farid Adilazuarda, Musa Izzanardi Wijanarko, Lucky Susanto, Khumaisa Nur’aini, Derry Tanti Wijaya, Alham Fikri Aji

Idea First, Code Later: Disentangling Problem Solving from Code Generation in Evaluating LLMs for Competitive Programming (2026.findings-acl)

Challenge:	NusaAksara covers 8 scripts across 7 languages, including low-resource languages not commonly seen in NLP benchmarks.
Approach:	They propose a benchmark for Indonesian scripts that includes their original scripts and a dataset that includes 8 scripts across 7 languages.
Outcome:	The proposed benchmark covers 8 scripts across 7 languages, including low-resource languages not commonly seen in NLP benchmarks.

Copied to clipboard

Sama Hadhoud, Alaa Elsetohy, Frederikus Hudi, Jan Christian Blaise Cruz, Steven Halim, Alham Fikri Aji

Programming Languages in NLP

Challenge:	Existing evaluations conflate algorithmic reasoning with code-level implementation.
Approach:	They propose to center editorials in both solution generation and evaluation . they propose to compare editorials to gold standards and validate an LLM-as-a-judge protocol .
Outcome:	The proposed approach improves solve rates on some LLMs with gold editorials . but the gap between gold and generated editorials shows bottlenecks in implementation .

From Multiple-Choice to Extractive QA: A Case Study for English and Arabic (2025.coling-main)

Copied to clipboard

Teresa Lynn, Malik H. Altakrori, Samar M. Magdy, Rocktim Jyoti Das, Chenyang Lyu, Mohamed Nasr, Younes Samih, Kirill Chirkunov, Alham Fikri Aji, Preslav Nakov, Shantanu Godbole, Salim Roukos, Radu Florian, Nizar Habash

Question Answering

Challenge:	Recent years have brought about very fast developments in Natural Language Processing (NLP), but many other languages are overlooked due to limited resources.
Approach:	They propose to repurpose a multilingual BELEBELE dataset for a task of extractive QA in the style of machine reading comprehension.
Outcome:	The proposed approach could be used to extract QA in the style of machine reading comprehension.

A Relation Extraction Dataset for Knowledge Extraction from Web Tables (2022.coling-1)

Copied to clipboard

Siffi Singh, Alham Fikri Aji, Gaurav Singh, Christos Christodoulopoulos

Information Extraction & Text Mining

Challenge:	Existing datasets with relational web-tables are either synthetic, or very small in size.
Approach:	They propose to annotate relational web-tables against a human-annotated dataset using crowd sourced annotators from MTurk.
Outcome:	The proposed dataset has 50x larger number of column pairs than the existing human-annotated benchmark.

MLKV: Multi-Layer Key-Value Heads for Memory Efficient Transformer Decoding (2025.findings-naacl)

Copied to clipboard

Zayd Muhammad Kawakibi Zuhri, Muhammad Farid Adilazuarda, Ayu Purwarianti, Alham Fikri Aji

BLOOM+1: Adding Language Support to BLOOM for Zero-Shot Prompting (2023.acl-long)

Challenge:	Multi-Layer Key-Value (MLKV) sharing reduces memory usage by 6x compared to Multi-Query Attention and Grouped-Query Attributes.
Approach:	They propose a novel approach that extends KV sharing across transformer layers to reduce memory usage beyond what was possible with Multi-Query Attention and Grouped-Query Attributes.
Outcome:	The proposed approach reduces KV cache size by 6x with minimal performance loss and scales linearly with model size, batch size, and sequence length.

Copied to clipboard

Zheng Xin Yong, Hailey Schoelkopf, Niklas Muennighoff, Alham Fikri Aji, David Ifeoluwa Adelani, Khalid Almubarak, M Saiful Bari, Lintang Sutawika, Jungo Kasai, Ahmed Baruwa, Genta Winata, Stella Biderman, Edward Raff, Dragomir Radev, Vassilina Nikoulina

Balanced Multi-Factor In-Context Learning for Multilingual Large Language Models (2025.emnlp-main)

Challenge:	Existing language adaptation strategies for multilingual models are limited to 46 languages . a new language is added to the model to improve zero-shot prompting performance .
Approach:	They apply existing language adaptation strategies to BLOOM and benchmark its zero-shot prompting performance on eight new languages in a resource-constrained setting.
Outcome:	The proposed model can be extended to other languages without incurring prohibitively large costs.

Copied to clipboard

Masahiro Kaneko, Alham Fikri Aji, Timothy Baldwin

The Decades Progress on Code-Switching Research in NLP: A Systematic Survey on Trends and Challenges (2023.findings-acl)

Challenge:	Existing approaches address key factors that influence multilingual ICL, but they do not integrate them into the model.
Approach:	They propose a method that quantifies and optimally balances three factors for improved example selection.
Outcome:	Experiments on mCSQA and TYDI show that the proposed method outperforms existing methods.

Copied to clipboard

Genta Winata, Alham Fikri Aji, Zheng Xin Yong, Thamar Solorio

A Paradigm Shift: The Future of Machine Translation Lies with Large Language Models (2024.lrec-main)

Challenge:	Code-Switching is a common phenomenon in written text and conversation . it is not so common to observe code-switching in spoken language and not in written language .
Approach:	They present a systematic survey on code-switching research in natural language processing to understand the progress of the past decades and conceptualize the challenges and tasks on the topic.
Outcome:	The proposed model combines linguistic theories and machine learning techniques to understand the code-switching phenomenon.

Copied to clipboard

Chenyang Lyu, Zefeng Du, Jitao Xu, Yitao Duan, Minghao Wu, Teresa Lynn, Alham Fikri Aji, Derek F. Wong, Longyue Wang

Style Over Substance: Evaluation Biases for Large Language Models (2025.coling-main)

Challenge:	Large Language Models (LLMs) are introducing a new phase in machine translation . despite advances in MT, there are still many challenges to overcome .
Approach:	They propose to highlight several new directions for MT that are influenced by Large Language Models like GPT-4 and ChatGPT.
Outcome:	The proposed models offer vast linguistic understandings and bring innovative methodologies, such as prompt-based techniques, that have the potential to further elevate MT.

Copied to clipboard

Minghao Wu, Alham Fikri Aji

On “Scientific Debt” in NLP: A Case for More Rigour in Language Model Pre-Training Research (2023.acl-long)

Challenge:	Ranking the relative performance of large language models based on Elo ratings is gaining popularity . however, the extent to which humans and LLMs are capable evaluators remains uncertain .
Approach:	They propose to evaluate machine-generated text across multiple dimensions using the Elo rating system . they propose to use crowd-sourced and expert annotators to rank models based on Elo ratings .
Outcome:	The proposed method improves the quality of LLM-based evaluations, but there is no improvement in crowd-sourced evaluations.

Copied to clipboard

Made Nindyatama Nityasya, Haryo Wibowo, Alham Fikri Aji, Genta Winata, Radityo Eko Prasojo, Phil Blunsom, Adhiguna Kuncoro