AI RESEARCH
When Context Sticks: Studying Interference in In-Context Learning
arXiv CS.LG
•
ArXi:2604.23371v1 Announce Type: new This paper investigates context stickiness in in-context learning (ICL), a phenomenon where earlier examples in a prompt interfere with a transformer's ability to adapt to later tasks. Using synthetic regression tasks over linear and quadratic functions, we examine how models trained under sequential, mixed, and random curricula handle abrupt task switches during inference.