\documentclass{beamer}
\begin{document}
\begin{frame}
\frametitle{Scope}
How to assess model capabilities.
\end{frame}
\begin{frame}
\frametitle{Metrics}
Perplexity, factuality, toxicity.
\end{frame}
\begin{frame}
\frametitle{Benchmarks}
MMLU, HellaSwag, TruthfulQA.
\end{frame}
\begin{frame}
\frametitle{Limitations}
Overfitting, brittleness, hallucinations.
\end{frame}
\end{document}

