Sensitivity of Small Language Models to Fine-tuning Data Contamination

Published in arXiv:2511.06763, 2025