Collection by 
Feb 11
-
A Unified View of Attention and Residual Sinks: Outlier-Driven Rescaling is Essential for Transformer Training
Paper • 2601.22966 • Published -
STEM: Scaling Transformers with Embedding Modules
Paper • 2601.10639 • Published • 2 -
Deep Delta Learning
Paper • 2601.00417 • Published • 34 -
mHC: Manifold-Constrained Hyper-Connections
Paper • 2512.24880 • Published • 331
