Giovana Kerche Bonás – Marcos M. Raimundo

Institute of Computing - University of Campinas

Publications

PRISM: Transfer and Limits of Activation Interventions in Tool-Calling Language Models

Giovana Kerche Bonás, T. Almeida, Marcelo Henrique da Silva Reis, Rodrigo Nogueira, Marcos M. Raimundo (2026). Preprint.

Combinatorial U-Curve Optimization for Multi-Task Learning: A Task Selection Approach

Giovana K Bonás, Marcelo S Reis, Marcos M Raimundo (2025). 2025 International Joint Conference on Neural Networks (IJCNN).

Projects

Mechanistic Interpretability and Latent Controllability in Large Language Models

Research on mechanistic interpretability, representation engineering (RepE), and attention steering in foundation models — investigating internal activation geometry to detect hallucinations, prevent scope drift, mitigate demographic bias, and enforce safe behavioral alignment without full-model retraining.