Xxxlia Lin Updated ^new^ 🔔

Alexia Lin, J. Chen, M. Rodriguez

The original Alexia Lin model (Lin et al., 2023) introduced a novel cross-attention module for few-shot learning. However, subsequent replication studies noted instability in high-variance support sets. This paper addresses those limitations with a comprehensive update, hence "Alexia Lin Updated."

We argue that "versioned" models like AL-V2 should become standard in FSL research, with clear changelogs and frozen checkpoints.