Research Explorer
Papers
Conferences
Authors
Topics
Keywords
Trends
Achievements
Explore
← Back to papers
2025
ICML
ICML 2025
The underlying structures of self-attention: symmetry, directionality, and emergent dynamics in Transformer training
Authors
Matteo Saponati
,
Pascal Josef Sager
,
Pau Vilimelis Aceituno
,
Thilo Stadelmann
,
Benjamin F Grewe
Download PDF
Related papers
Scaling Sparse Feature Circuits For Studying In-Context Learning
2025
Incremental Gradient Descent with Small Epoch Counts is Surprisingly Slow on Ill-Conditioned Problems
2025
SToFM: a Multi-scale Foundation Model for Spatial Transcriptomics
2025
Batch List-Decodable Linear Regression via Higher Moments
2025
GS-Bias: Global-Spatial Bias Learner for Single-Image Test-Time Adaptation of Vision-Language Models
2025