Gaël Guibon

Emotional information in LLMs is more interconnected than previously thought, with optimal probing layers varying significantly across different text sources.

Textual data from financial reports can significantly enhance fraud detection, as evidenced by our framework's superior performance on a novel benchmark task.

Ontology-driven constraints unlock modular, explainable control over LLM conversations, enabling targeted fine-tuning for specific aspects like English proficiency or content polarity.