Portrait of Till Speicher

Till Speicher

PhD Student

MPI-SWS

About me

I am a final-year PhD student at the Max Planck Institute for Software Systems (MPI-SWS), Germany, advised by Krishna Gummadi. I obtained a Bachelor's degree in Computer Science from Saarland University.

Currently, I am working on understanding how large language models learn, particularly how they memorize information, learn formal grammars, and acquire factual knowledge. Previously, I have worked on topics related to fairness in algorithmic decision-making, and on understanding the properties of representations in deep neural networks.

In my free time, I train for and compete in triathlons 🏊‍♂️🚴‍♂️🏃‍♂️.

You can find more information in my CV.

Research

Understanding Memorisation in LLMs: Dynamics, Influencing Factors, and Implications

Till Speicher, Mohammad Aflah Khan, Qinyuan Wu, Vedant Nanda, Soumi Das, Bishwamittra Ghosh, Krishna P. Gummadi, Evimaria Terzi

Preprint, 2024

Paper

Towards Reliable Latent Knowledge Estimation in LLMs: In-Context Learning vs. Prompting Based Factual Knowledge Extraction

Qinyuan Wu, Mohammad Aflah Khan, Soumi Das, Vedant Nanda, Bishwamittra Ghosh, Camila Kolling, Till Speicher, Laurent Bindschaedler, Krishna P. Gummadi, Evimaria Terzi

Preprint, 2024

Paper

Understanding the Role of Invariance in Transfer Learning

Till Speicher, Vedant Nanda, Krishna P. Gummadi

TMLR 2024

PaperCode

Diffused Redundancy in Pre-trained Representations

Vedant Nanda, Till Speicher, John P. Dickerson, Soheil Feizi, Krishna P. Gummadi, Adrian Weller

NeurIPS 2023

PaperCode

Pointwise Representational Similarity

Camila Kolling, Till Speicher, Vedant Nanda, Mariya Toneva, Krishna P. Gummadi

Preprint, 2023

Paper

Measuring Representational Robustness of Neural Networks Through Shared Invariances

Vedant Nanda, Till Speicher, Camila Kolling, John P. Dickerson, Krishna P. Gummadi, Adrian Weller

ICML 2022, oral presentation

PaperCode

Unifying Model Explainability and Robustness via Machine-Checkable Concepts

Vedant Nanda, Till Speicher, John P. Dickerson, Krishna P. Gummadi, Muhammad Bilal Zafar

Preprint, 2020

Paper

A Unified Approach to Quantifying Algorithmic Unfairness: Measuring Individual & Group Unfairness via Inequality Indices

Till Speicher, Hoda Heidari, Nina Grgic-Hlaca, Krishna P. Gummadi, Adish Singla, Adrian Weller, Muhammad Bilal Zafar

KDD 2018

PaperCode

Potential for Discrimination in Online Targeted Advertising

Till Speicher, Muhammad Ali, Giridhari Venkatadri, Filipe Nunes Ribeiro, George Arvanitakis, Fabrício Benevenuto, Krishna P. Gummadi, Patrick Loiseau, Alan Mislove

FAccT 2018

Paper

Reliable Learning by Subsuming a Trusted Model: Safe Exploration of the Space of Complex Models

Till Speicher, Muhammad Bilal Zafar, Krishna P. Gummadi, Adish Singla, Adrian Weller

Reliable Machine Learning in the Wild - ICML Workshop 2017

Paper

A Generalized Language Model as the Combination of Skipped n-grams and Modified Kneser-Ney Smoothing

Rene Pickhardt, Thomas Gottron, Martin Körner, Paul Georg Wagner, Till Speicher, Steffen Staab

ACL 2024

PaperCode