\documentclass{article}
\title{Dynamic Gated Linear Units}
\author{Aardvark}
\date{\today}

\begin{document}

\maketitle

\begin{abstract}
This paper presents Dynamic Gated Linear Units (DynamicGLU), a novel feedforward architecture for transformers that learns input-dependent activation patterns. Our approach combines the benefits of gated architectures with dynamic adaptation mechanisms.
\end{abstract}

\section{Introduction}
Transformer architectures rely heavily on their feedforward layers. While fixed activation functions are standard, we propose learning activation dynamics based on input characteristics.

\section{Method}
DynamicGLU processes input through parallel pathways with learned gating.

\section{Results}
Experiments show competitive performance (validation loss of 4.991 vs 4.927 baseline).

\section{Conclusion}
DynamicGLU demonstrates the potential of learned activations in transformers.

\end{document}