Tag
This paper proposes improvements to Relative Representations by learning robust semantic anchors and using a geometry-aware similarity metric, enabling nearly lossless information transfer and stable zero-shot communication between independently trained models of varying architectures.