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...
Read More →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.
Uncovering how neural networks process and represent language, making AI systems more transparent and trustworthy through mechanistic interpretability.
Developing efficient methods for training models with minimal labeled data, reducing annotation costs while maintaining high performance.
Analyzing and understanding the behavior of large-scale language models, exploring their capabilities and limitations.
Creating intelligent systems for condensing scientific papers and documents while preserving critical information and context.
How do we identify which parts of a neural network are responsible for specific behaviors? Exploring mechanistic interpretability techniques...
Read More →Why label millions of examples when you can achieve similar performance with thousands? A practical guide to active learning strategies...
Read More →Investigating how large language models store and retrieve factual knowledge. Surprising findings about relation-specific neurons...
Read More →Reflections on navigating research challenges, finding your niche, and balancing work with life as an ELLIS PhD student...
Read More →Interested in collaborating on interpretability or active learning? Let's push the boundaries of NLP together.