The Shape of Learning: Anisotropy and Intrinsic Dimensions in Transformer-Based Models
This provides incremental insights into embedding properties for researchers working on transformer architectures.
The study investigated anisotropy dynamics and intrinsic dimensions in transformer embeddings, revealing that decoders exhibit a bell-shaped anisotropy curve peaking in middle layers, unlike encoders, and that intrinsic dimensions expand early in training then compress later.
In this study, we present an investigation into the anisotropy dynamics and intrinsic dimension of embeddings in transformer architectures, focusing on the dichotomy between encoders and decoders. Our findings reveal that the anisotropy profile in transformer decoders exhibits a distinct bell-shaped curve, with the highest anisotropy concentrations in the middle layers. This pattern diverges from the more uniformly distributed anisotropy observed in encoders. In addition, we found that the intrinsic dimension of embeddings increases in the initial phases of training, indicating an expansion into higher-dimensional space. Which is then followed by a compression phase towards the end of training with dimensionality decrease, suggesting a refinement into more compact representations. Our results provide fresh insights to the understanding of encoders and decoders embedding properties.