4 papers
How far can bias go? Tracing bias from pretraining data to alignment
Marion Thaler, Abdullatif Köksal, Alina Leidinger +2
As LLMs are increasingly integrated into user-facing applications, addressing biases that perpetuate societal inequalities is crucial. While much work has gone into measuring or mi…
SYNTHEVAL: Hybrid Behavioral Testing of NLP Models with Synthetic CheckLists
Raoyuan Zhao, Abdullatif Köksal, Yihong Liu +3
Traditional benchmarking in NLP typically involves using static held-out test sets. However, this approach often results in an overestimation of performance and lacks the ability t…
TurkishMMLU: Measuring Massive Multitask Language Understanding in Turkish
Arda Yüksel, Abdullatif Köksal, Lütfi Kerem Åenel +2
Multiple choice question answering tasks evaluate the reasoning, comprehension, and mathematical abilities of Large Language Models (LLMs). While existing benchmarks employ automat…
MURI: High-Quality Instruction Tuning Datasets for Low-Resource Languages via Reverse Instructions
Abdullatif Köksal, Marion Thaler, Ayyoob Imani +3
Instruction tuning enhances large language models (LLMs) by aligning them with human preferences across diverse tasks. Traditional approaches to create instruction tuning datasets…