García-Carrasco, J.; Maté, A.; Trujillo, J. Extracting Interpretable Task-Specific Circuits from Large Language Models for Faster Inference. AAAI 2025, 39, 16772-16780.