Sebastian Koralewski

Sebastian Koralewski, Merwan Barlier, Yulia Stolin et al.

Online recommendation platforms increasingly use Large Language Models (LLMs) to extract structured features from ad creatives. While deploying a single-call LLM annotation agent yields significant Click-Through Rate (CTR) improvements in our live production environment, per-creative prompting is prohibitively expensive to scale. The redundant system instructions sent in every request account for 94% of billed input tokens. To break this cost bottleneck, we introduce PromptPack, a scalable, high-throughput LLM annotation agent. PromptPack achieves this scale via in-context batching, combining a shared system prompt, a strict XML structural envelope, and an output correction layer to ensure deterministic, pipeline-ready feature extraction across multiple creatives simultaneously. We evaluate PromptPack via an offline retrieval benchmark using a downstream logistic-regression ranker. To deeply profile the agent's behavior, we measure AUC and introduce Volume-Weighted Absolute Lift (VWAL), a novel metric capturing the signal quality of the generated features. Compared to our live, unbatched production baseline, PromptPack at batch size 20 cuts our LLM costs by 89% and accelerates throughput by 2.5x while fully preserving AUC.

Michael Neumann, Sebastian Koralewski, Michael Beetz

Anticipating what might happen as a result of an action is an essential ability humans have in order to perform tasks effectively. On the other hand, robots capabilities in this regard are quite lacking. While machine learning is used to increase the ability of prospection it is still limiting for novel situations. A possibility to improve the prospection ability of robots is through simulation of imagined motions and the physical results of these actions. Therefore, we present URoboSim, a robot simulator that allows robots to perform tasks as mental simulation before performing this task in reality. We show the capabilities of URoboSim in form of mental simulations, generating data for machine learning and the usage as belief state for a real robot.

Gayane Kazhoyan, Simon Stelter, Franklin Kenghagho Kenfack et al.

In this paper, we present an experiment, designed to investigate and evaluate the scalability and the robustness aspects of mobile manipulation. The experiment involves performing variations of mobile pick and place actions and opening/closing environment containers in a human household. The robot is expected to act completely autonomously for extended periods of time. We discuss the scientific challenges raised by the experiment as well as present our robotic system that can address these challenges and successfully perform all the tasks of the experiment. We present empirical results and the lessons learned as well as discuss where we hit limitations.