mlps

5 papers

• Beyond Components: Singular Vector-Based Interpretability of Transformer Circuits
• EUGens: Efficient, Unified and General Dense Layers
• From Kolmogorov to Cauchy: Shallow XNet Surpasses KANs
• GeRaF: Neural Geometry Reconstruction from Radio Frequency Signals
• U-REPA: Aligning Diffusion U-Nets to ViTs