4 papers
Assessment of RAG and Fine-Tuning for Industrial Question-Answering-Applications
Jakob Sturm, Josef Pichlmeier, Christian Bernhard +4
Large Language Models (LLMs) are increasingly employed in enterprise question-answering (QA) systems, requiring adaptation to domain-specific knowledge. Among the most prevalent me…
XL-SafetyBench: A Country-Grounded Cross-Cultural Benchmark for LLM Safety and Cultural Sensitivity
Dasol Choi, Eugenia Kim, Jaewon Noh +14
Current LLM safety benchmarks are predominantly English-centric and often rely on translation, failing to capture country-specific harms. Moreover, they rarely evaluate a model's a…
Harnessing Chain-of-Thought Metadata for Task Routing and Adversarial Prompt Detection
Ryan Marinelli, Josef Pichlmeier, Tamas Bisztray
In this work, we propose a metric called Number of Thoughts (NofT) to determine the difficulty of tasks pre-prompting and support Large Language Models (LLMs) in production context…
Performance Characterization of Expert Router for Scalable LLM Inference
Josef Pichlmeier, Philipp Ross, Andre Luckow
Large Language Models (LLMs) have experienced widespread adoption across scientific and industrial domains due to their versatility and utility for diverse tasks. Nevertheless, dep…