publications

2026

1. arXiv 2026

   Write Once, Run Everywhere: The Axon DSL for Shape-Safe and Framework-Agnostic LLM Architectures

   Jacob Nielsen, Danial Namazifard, Lukas Galke Poech, and Peter Schneider-Kamp

   arXiv preprint arXiv:2608.19889, Aug 2026

   Abs arXiv

   We present Axon, a strongly typed domain-specific language for defining large language model architectures independently of a particular machine learning framework. Axon specifications can be compiled into standalone implementations for PyTorch, PyTorch with Triton, JAX, MLX, and vLLM, enabling the same model architecture to be deployed across different training and inference environments. Extensive inference benchmarks across models ranging from 135M to 32B parameters show that generated implementations can match or substantially outperform reference Transformers implementations across several backends, while providing concise and auditable model specifications.

2. SilkRoadNLP 2026

   APARSIN: A Multi-Variety Sentiment and Translation Benchmark for Iranic Languages

   Sadegh Jafari, Tara Azin, Farhad Roodi, Zahra Dehghani Tafti, Mehrdad Ghadrdan, Elham Vatankhahan Esfahani, Aylin Naebzadeh, Mohammadhadi Shahhosseini, Ghafoor Khan, Kazem Forghani, Danial Namazi, Seyed Mohammad Hossein Hashemi, Farhan Farsi, Mohammad Osoolian, Maede Mohammadi, Mohammad Erfan Zare, Muhammad Hasnain Khan, Muhammad Hussain, Nooreen Zaki, Joma Mohammadi, Shayan Bali, Mohammad Javad Ranjbar, Els Lefever, and Veronique Hoste

   In The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family, Mar 2026

   Abs ACL Code

   The Iranic language family includes many underrepresented languages and dialects that remain largely unexplored in modern NLP research. We introduce APARSIN, a multi-variety benchmark covering 14 Iranic languages, dialects, and accents, designed for sentiment analysis and machine translation. The dataset includes both high- and low-resource varieties, several of which are endangered, capturing linguistic variation across them. We evaluate a set of instruction-tuned Large Language Models (LLMs) on these tasks and analyze their performance across the varieties. Our results highlight substantial performance gaps between standard Persian and other Iranic languages and dialects, demonstrating the need for more inclusive multilingual and dialectally diverse NLP benchmarks.

3. SilkRoadNLP 2026

   A Computational Approach to Language Contact – A Case Study of Persian

   Ali Basirat, Danial Namazifard, and Navid Baradaran Hemmati

   In The Proceedings of the First Workshop on NLP and LLMs for the Iranian Language Family, Mar 2026

   Abs ACL arXiv

   We investigate structural traces of language contact in the intermediate representations of a monolingual language model. Focusing on Persian as a historically contact-rich language, we probe the representations of a Persian-trained model when exposed to languages with varying degrees and types of contact with Persian. Our methodology quantifies the amount of linguistic information encoded in intermediate representations and assesses how this information is distributed across model components for different morphosyntactic features. The results show that universal syntactic information is largely insensitive to historical contact, whereas morphological features such as CASE and GENDER are strongly shaped by language-specific structure, suggesting that contact effects in monolingual language models are selective and structurally constrained.

2025

1. IJCNLP-AACL 2025

   Isolating Culture Neurons in Multilingual Large Language Models

   Danial Namazifard and Lukas Galke Poech

   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, Dec 2025

   Abs ACL arXiv Code

   Language and culture are deeply intertwined, yet it has been unclear how and where multilingual large language models encode culture. Here, we build on an established methodology for identifying language-specific neurons to localize and isolate culture-specific neurons, carefully disentangling their overlap and interaction with language-specific neurons. To facilitate our experiments, we introduce MUREL, a curated dataset of 85.2 million tokens spanning six different cultures. Our localization and intervention experiments show that LLMs encode different cultures in distinct neuron populations, predominantly in upper layers, and that these culture neurons can be modulated largely independently of language-specific neurons or those specific to other cultures. These findings suggest that cultural knowledge and propensities in multilingual language models can be selectively isolated and edited, with implications for fairness, inclusivity, and alignment.

2. COLING 2025

   Comparative Study of Multilingual Idioms and Similes in Large Language Models

   Paria Khoshtab, Danial Namazifard, Mostafa Masoudi, Ali Akhgary, Samin Mahdizadeh Sani, and Yadollah Yaghoobzadeh

   In Proceedings of the 31st International Conference on Computational Linguistics, Jan 2025

   Abs ACL arXiv Code

   This study addresses the gap in the literature concerning the comparative performance of LLMs in interpreting different types of figurative language across multiple languages. By evaluating LLMs using two multilingual datasets on simile and idiom interpretation, we explore the effectiveness of various prompt engineering strategies, including chain-of-thought, few-shot, and English translation prompts. We extend the language of these datasets to Persian as well by building two new evaluation sets. Our comprehensive assessment involves both closed-source models and open-source models, highlighting significant differences in performance across languages and figurative types. Our findings reveal that while prompt engineering methods are generally effective, their success varies by figurative type, language, and model. We also observe that open-source models struggle particularly with low-resource languages in similes. Additionally, idiom interpretation is nearing saturation for many languages, necessitating more challenging evaluations.

2024

1. LREC-COLING 2024

   Benchmarking Large Language Models for Persian: A Preliminary Study Focusing on ChatGPT

   Amirhossein Abaskohi, Sara Baruni, Mostafa Masoudi, Nesa Abbasi, Mohammad Hadi Babalou, Ali Edalat, Sepehr Kamahi, Samin Mahdizadeh Sani, Nikoo Naghavian, Danial Namazifard, Pouya Sadeghi, and Yadollah Yaghoobzadeh

   In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), May 2024

   Abs ACL arXiv Code

   This paper explores the efficacy of large language models (LLMs) for Persian. While ChatGPT and consequent LLMs have shown remarkable performance in English, their efficiency for more low-resource languages remains an open question. We present the first comprehensive benchmarking study of LLMs across diverse Persian language tasks. Our primary focus is on GPT-3.5-turbo, but we also include GPT-4 and OpenChat-3.5 to provide a more holistic evaluation. Our assessment encompasses a diverse set of tasks categorized into classic, reasoning, and knowledge-based domains. To enable a thorough comparison, we evaluate LLMs against existing task-specific fine-tuned models. Given the limited availability of Persian datasets for reasoning tasks, we introduce two new benchmarks: one based on elementary school math questions and another derived from the entrance exams for 7th and 10th grades. Our findings reveal that while LLMs, especially GPT-4, excel in tasks requiring reasoning abilities and a broad understanding of general knowledge, they often lag behind smaller pretrained models fine-tuned specifically for particular tasks. Additionally, we observe improved performance when test sets are translated to English before inputting them into GPT-3.5. These results highlight the significant potential for enhancing LLM performance in the Persian language.