1 paper across 1 session
Representational alignment based on concept discovery across ViTs trained on different tasks.