Context, Models and Prompt Optimization for Automated Hallucination Detection in LLM Output
Our winning system for the SemEval-2025 Mu-SHROOM shared task, which ranked #1 on average across 14 languages for pinpointing hallucinated text in LLM outputs.