Ahmad Dawar Hakimi

ELLIS PhD Student

CIS @ LMU Munich · CopeNLU

Pushing the boundaries of Natural Language Processing through interpretability research, active learning, and large language model analysis. Co-supervised by Hinrich Schütze and Isabelle Augenstein.

View Publications Get in Touch

Research Focus

🔍 Mechanistic Interpretability Research

Uncovering how neural networks process and represent language, making AI systems more transparent and trustworthy through mechanistic interpretability.

🎯 Active Learning

Developing efficient methods for training models with minimal labeled data, reducing annotation costs while maintaining high performance.

🤖 Large Language Models

Analyzing and understanding the behavior of large-scale language models, exploring their capabilities and limitations.

Publications

On relation-specific neurons in large language models

Liu, Yihong, Chen, Runsheng, Hirlimann, Lea, Hakimi, Ahmad Dawar, Wang, Mingyang, Kargaran, Amir Hossein, Rothe, Sascha, Yvon, François, Schütze, Hinrich

arXiv preprint arXiv:2502.17355, 2025

Time Course MechInterp: Analyzing the Evolution of Components and Knowledge in Large Language Models

Hakimi, Ahmad Dawar, Modarressi, Ali, Wicke, Philipp, Schütze, Hinrich

arXiv preprint arXiv:2506.03434, 2025

BlackboxNLP-2025 MIB Shared Task: Exploring Ensemble Strategies for Circuit Localization Methods

Mondorf, Philipp, Wang, Mingyang, Gerstner, Sebastian, Hakimi, Ahmad Dawar, Liu, Yihong, Veloso, Leonor, Zhou, Shijia, Schütze, Hinrich, Plank, Barbara

Proceedings of the 8th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP: 537-542, 2025

Citance-contextualized summarization of scientific papers

Syed, Shahbaz, Hakimi, Ahmad Dawar, Al Khatib, Khalid, Potthast, Martin

Findings of the Association for Computational Linguistics: EMNLP 2023: 8551-8568, 2023

Explorative visual analysis of rap music

Meinecke, Christofer, Hakimi, Ahmad Dawar, Jänicke, Stefan

Information 13(1): 10, 2021

Casting the same sentiment classification problem

Körner, Erik, Hakimi, Ahmad Dawar, Heyer, Gerhard, Potthast, Martin

Findings of the Association for Computational Linguistics: EMNLP 2021: 584-590, 2021

On classifying whether two texts are on the same side of an argument

Körner, Erik, Wiedemann, Gregor, Hakimi, Ahmad Dawar, Heyer, Gerhard, Potthast, Martin

Proceedings of the 2021 conference on empirical methods in natural language processing: 10130-10138, 2021

The road map to FAME: A framework for mining and formal evaluation of arguments

Baumann, Ringo, Wiedemann, Gregor, Heinrich, Maximilian, Hakimi, Ahmad Dawar, Heyer, Gerhard

Datenbank-Spektrum 20(2): 107-113, 2020

Latest Blog Posts

🧠

Dec 20, 2024 • Interpretability

Understanding Neural Circuits: A Deep Dive

How do we identify which parts of a neural network are responsible for specific behaviors? Exploring mechanistic interpretability techniques...

🎯

Nov 15, 2024 • Active Learning

Efficient Training with Active Learning

Why label millions of examples when you can achieve similar performance with thousands? A practical guide to active learning strategies...

🤖

Oct 28, 2024 • LLMs

What Do Language Models Really Know?

Investigating how large language models store and retrieve factual knowledge. Surprising findings about relation-specific neurons...

📚

Sep 10, 2024 • Research Life

Lessons from My First Year as a PhD Student

Reflections on navigating research challenges, finding your niche, and balancing work with life as an ELLIS PhD student...