Ahmad Dawar Hakimi

ELLIS PhD Student
CIS @ LMU Munich · CopeNLU @ UCPH

Hi, I’m Dawar, a second-year ELLIS PhD student at LMU Munich and the University of Copenhagen, co-supervised by Hinrich Schütze and Isabelle Augenstein. In my research, I work on training dynamics in large language models and active learning in the era of LLMs. When I’m not in the lab studying training dynamics in llms, I’m working on training dynamics in humans with Early Birds Urban Movement or enjoying some beautiful music .

View Publications Get in Touch
Ahmad Dawar Hakimi
Interpretability Active Learning LLMs

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

Showing publications related to:

On Relation-Specific Neurons in Large Language Models

Yihong Liu*, Runsheng Chen*, Lea Hirlimann, Ahmad Dawar Hakimi, Mingyang Wang, Amir Hossein Kargaran, Sascha Rothe, François Yvon, Hinrich Schütze
EMNLP 2025
interpretability llms

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

Ahmad Dawar Hakimi, Ali Modarressi, Philipp Wicke, Hinrich Schütze
Findings of ACL 2025
interpretability llms

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

Philipp Mondorf, Mingyang Wang, Sebastian Gerstner, Ahmad Dawar Hakimi, Yihong Liu, Leonor Veloso, Shijia Zhou, Hinrich Schütze, Barbara Plank
BlackboxNLP Workshop @ EMNLP 2025
interpretability llms

Citance-Contextualized Summarization of Scientific Papers

Shahbaz Syed*, Ahmad Dawar Hakimi*, Khalid Al-Khatib, Martin Potthast
Findings of EMNLP 2023
summarization

Explorative Visual Analysis of Rap Music

Christofer Meinecke, Ahmad Dawar Hakimi, Stefan Jänicke
Information, 13(1), 2022
visualization

Casting the Same Sentiment Classification Problem

Erik Körner, Ahmad Dawar Hakimi, Gerhard Heyer, Martin Potthast
Findings of EMNLP 2021
sentiment

On Classifying Whether Two Texts Are on the Same Side of an Argument

Erik Körner, Gregor Wiedemann, Ahmad Dawar Hakimi, Gerhard Heyer, Martin Potthast
EMNLP 2021
argument-mining

The Road Map to FAME: A Framework for Mining and Formal Evaluation of Arguments

Ringo Baumann, Gregor Wiedemann, Maximilian Heinrich, Ahmad Dawar Hakimi, Gerhard Heyer
Datenbank-Spektrum, 20(2): 107–113, 2020
argument-mining

Blog

✍️

Coming Soon

I'm working on writing up thoughts on mechanistic interpretability, active learning, and life as a PhD student. Stay tuned!

Let's Connect

Interested in collaborating on interpretability or active learning? Let's push the boundaries of NLP together.