👋 Hey, I'm Lasse
A computer scientist with a background in economics specializing in the interpretability and reliability of Large Language Models. My research focuses on improving language models as reliable, large-scale text processing tools by leveraging the models internals for information extraction and task adaptation. Through my background in economics I ground my research not only in technical facination, but also in the social and economical implications of more reliable AI technology.
Please contact me at lassejantsch [at] web.de.
📄 Publications
A Metric-Based Direct Preference Optimization Method for Human-Aligned Summarization
Seonghwan Yoon, Lasse Jantsch, Hwansong Joo, Young-Kyoon Suh
Korean Computer Congress (KCC) 2025
QRiousRAG: Retrieval Augmented Generation with Question-refinement using LLMs in ambiguous question answering
Dong-Jae Koh, Lasse Jantsch, Hwansong Joo, Young-Kyoon Suh
Korean Software Congress (KSC) 2024
🏆 Best Poster Award
SciConBERT: A Dynamic Citation Context Extraction Model Using Siamese BERT
Lasse Jantsch, Jisu Lee, Joel Thomas, Jiarun Cheong, Young-Kyoon Suh
Korean Computer Congress (KCC) 2024
Analyzing the Impact of Rising Inequality and the Covid-19 Pandemic on News Consumption Patterns Across Different Demographics
Lasse Jantsch, Sien Papousek, Hyejung Shin, Anna-Katarina Kusimäki, Pauline Le Marrec, Chankyung Pak
The 11th Korea Media Panel Academic Conference 2023
🎓 Education
M.Sc. Computer Science - Kyungpook National University
Focus: Natural Language Processing, Deep Learning
2024 - 2026
B.A. Computer Science - University of Hamburg
Deep Learning for Sovereign Risk Contagion Modelling
2020 - 2024