Sadegh Mohammadian

I am a final-year undergraduate student in Computer Engineering at Sharif University of Technology.

I am broadly interested in model architecture design and sequence modeling, as well as reasoning and interpretability in vision-language models. I was a research intern at the EPFL NeuroAI Lab, where I worked with Prof. Martin Schrimpf. I also worked with Ali Behrouz at Cornell University on memory-based architectures for long video generation. I have worked with Prof. Mahdieh Soleymani at the MLL Lab on reasoning and interpretability in vision-language models, and was a research intern at the Qatar Computing Research Institute (QCRI) with Prof. Ehsaneddin Asgari.

Email / Google Scholar / Github / LinkedIn

Publications

	Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding Hosein Hasani, Amirmohammad Izadi, Fatemeh Askari, Mobin Bagherian, Sadegh Mohammadian*, Mohammad Izadi, Mahdieh Soleymani Baghshah ICML, 2026 arXiv / code
	Understanding Counting Mechanisms in Large Language and Vision-Language Models Hosein Hasani, Amirmohammad Izadi, Fatemeh Askari, Mobin Bagherian, Sadegh Mohammadian*, Mohammad Izadi, Mahdieh Soleymani Baghshah CVPR, 2026 arXiv / code
	Mechanistic Interpretability of Large-Scale Counting in LLMs through a System-2 Strategy Hosein Hasani, Mohammadali Banayeeanzade, Ali Nafisi, Sadegh Mohammadian, Fatemeh Askari, Mobin Bagherian, Amirmohammad Izadi, Mahdieh Soleymani Baghshah Findings of ACL, 2026 arXiv /
	EYE-Q: A Multilingual Benchmark for Visual Word Puzzle Solving and Image-to-Phrase Reasoning Ali Najar, Alireza Mirrokni, Arshia Izadyari, Sadegh Mohammadian, Amir Homayoon Sharifizade, Asal Meskin, Mobin Bagherian*, Ehsaneddin Asgari Under Review, 2026 arXiv / code / dataset

Research Experiences

NeuroAI Lab Vision Language Models • Jul. 2026 to Now Advisor: Prof. Martin Schrimpf
QCRI LLM Lab Vision Language Models • Nov. 2025 to Now Advisor: Prof. Ehsaneddin Asgari
Sharif ML Lab Vision Language Models • Apr. 2025 to Now Advisor: Prof. Mahdieh Soleymani

Design and source code from Jon Barron's website

Publications

Uncovering Grounding IDs: How External Cues Shape Multi-Modal Binding

Understanding Counting Mechanisms in Large Language and Vision-Language Models

Mechanistic Interpretability of Large-Scale Counting in LLMs through a System-2 Strategy

EYE-Q: A Multilingual Benchmark for Visual Word Puzzle Solving and Image-to-Phrase Reasoning

Research Experiences