data/tables/dev/val_tab_0022.tex

\documentclass[varwidth=598pt]{standalone}

\usepackage{booktabs}      % professional-quality tables
\usepackage{multirow}      % multi-row cells in tables
\usepackage{colortbl}      % color for tables (\cellcolor, \rowcolor)
\usepackage[table]{xcolor} % enhanced colors for tables
\usepackage{array}         % more flexible column formats (often used)
\usepackage{tabularx}      % for tables with auto-stretch columns (if used)
\usepackage{graphicx}      % for images in tables or figure/table floats
\usepackage{amssymb}       % math symbols
\usepackage{amsmath}       % math environments
\usepackage{soul}          % highlighting (used for colored text/cells)
\usepackage[normalem]{ulem} % underlining, strikethroughs
\usepackage[T1]{fontenc}   % font encoding
\usepackage[utf8]{inputenc}% input encoding (legacy, fine for pdflatex)
\usepackage{microtype}     % better text appearance
\usepackage{hyperref}      % hyperlinks
\usepackage{textcomp}      % for extra symbols
\usepackage{enumitem}      % for compact lists (if used in table notes)
\usepackage{adjustbox}
\usepackage{tabu}
\usepackage{pifont}        % http://ctan.org/pkg/pifont
\usepackage{bbding}        % \XSolidBrush
\usepackage{makecell}

\begingroup
\makeatletter
\renewcommand{\fnum@table}{}%
\long\def\@makecaption#1#2{%
  \vskip\abovecaptionskip
  \centering #2\par
  \vskip\belowcaptionskip
}
\makeatother


\begin{document}
\begin{table}
\centering
\begin{adjustbox}{max width=\textwidth}
    \begin{tabular}{@{}*{6}{l}@{}}
        & & {\psnrlabel} & {\ssimlabel} & {\lpipslabel} & {\paramslabel} \\
        \toprule
        \multirow{7}{*}{\rot{\textbf{Hidden dim}}}
        & 8 & 25.61 & 0.898 & 0.086 & 86.3K \\
        & 16 & 26.11 & 0.909 & 0.070 & 88.4K \\
        & \textit{32} & 26.53 & 0.913 & \underline{0.066} & 92.5K \\
        & 64 & 26.53 & 0.913 & \underline{0.066} & 100.7K \\
        & 128 & 26.51 & 0.912 & \underline{0.066} & 117.2K \\
        & 256 & \underline{26.77} & \underline{0.915} & \textbf{0.063} & 150.1K \\
        & 512 & \textbf{26.78} & \textbf{0.916} & \textbf{0.063} & 215.9K \\
        \midrule[\heavyrulewidth]
        \multirow{7}{*}{\rot{\textbf{Embed dim}}}
        & 8\ts\textdagger & 21.67 & 0.814 & 0.258 & 2.0K \\
        & 16\ts\textdagger & 23.22 & 0.853 & 0.183 & 4.5K \\
        & 32\ts\textdagger & 24.94 & 0.875 & 0.110 & 10.9K \\
        & 64\ts\textdagger & 25.69 & 0.898 & 0.084 & 29.9K \\
        & \textit{128}\ts\textdagger & \underline{26.05} & \underline{0.904} & \underline{0.075} & 92.5K \\
        & 256\ts\textdagger & \textbf{26.36} & \textbf{0.911} & \textbf{0.067} & 316.0K \\
        & 512\ts\textdagger & 19.93 & 0.768 & 0.274 & 1.2M \\
        \midrule[\heavyrulewidth]
        \multirow{6}{*}{\rot{\textbf{Heads}}}
        & 1 & 25.01 & 0.890 & 0.096 & 76.0K \\
        & \textit{4} & 26.53 & 0.913 & 0.066 & 92.5K \\
        & 8 & 26.77 & 0.916 & 0.062 & 92.5K \\
        & 16 & 26.78 & 0.917 & 0.062 & 92.5K \\
        & 32 & \textbf{27.01} & \textbf{0.920} & \textbf{0.060} & 92.5K \\
        & 64 & \underline{26.94} & \underline{0.919} & \underline{0.061} & 92.5K \\
        \midrule[\heavyrulewidth]
        \multirow{6}{*}{\rot{\textbf{Depth}}}
        & \emph{ViTCA--1} & \textbf{26.53} & \textbf{0.913} & \textbf{0.066} & 92.5K \\
        & ViTCA--2\ts\textdagger & \underline{10.82} & \underline{0.225} & \underline{0.771} & 175.3K \\
        & ViTCA--3\ts\textdagger & 9.70 & 0.165 & 0.793 & 258.0K \\
        \cmidrule(l){2-6}
        & \emph{ViT--1} & 19.70 & 0.779 & 0.237 & 83.9K \\
        & ViT--2\ts\textdagger & \underline{25.20} & \underline{0.900} & \underline{0.074} & 166.7K \\
        & ViT--3\ts\textdagger & \textbf{26.10} & \textbf{0.914} & \textbf{0.065} & 249.4K \\
        \midrule[\heavyrulewidth]
        \multirow{6}{*}{\rot{\textbf{PE type}}}
        & sincos5 & \underline{26.92} & \underline{0.917} & \underline{0.062} & 95.1K \\
        & sincos5xy & \textbf{27.00} & \textbf{0.919} & \textbf{0.059} & 95.3K \\
        & xy & 26.45 & 0.894 & 0.077 & 92.8K \\
        & \textit{handcrafted} & 26.53 & 0.913 & 0.066 & 92.5K \\
        & learned & 26.16 & 0.910 & 0.071 & 223.6K \\
        & none & 26.28 & 0.890 & 0.081 & 92.5K \\
        \bottomrule
        &&&&&\\ 
    \end{tabular}
\end{adjustbox}
\caption{Table 2: Quantitative ablation for denoising autoencoding with ViTCA (unless otherwise stated via prefix) on CelebA liu2015faceattributes. Boldface and underlining denote best and second best results. Italicized items denote baseline configuration settings. \ts\textdagger Trained with gradient checkpointing chen2016training, which slightly alters round-off error during backpropagation, resulting in slight variations of results compared to training without checkpointing. See Appendix subsec:extended:ablation.}
\end{table}
\end{document}