code/report/tables/training.tex

\begin{tabular}{lrrrrr}
\toprule
variant & $\alpha$ & $\lambda_{\mathrm{sup}}$ & $\lambda_h$ & val $d_H$ & val $d_q$ \\
\midrule
terminal-only ($\alpha=0$) & 0.00 & 0.01 & -- & 0.0130 & 0.1989 \\
original & 0.05 & 0.01 & -- & 0.0147 & 0.0555$^\dagger$ \\
no support ($\lambda_{\mathrm{sup}}=0$) & 0.05 & 0.00 & -- & 0.0146 & 0.0556 \\
arrival ($\lambda_h=0$) & 0.05 & 0.01 & 0.0 & 0.0263 & 0.0125 \\
arrival+hold ($\lambda_h=1$) & 0.05 & 0.01 & 1.0 & 0.0195 & 0.0130 \\
\textbf{arrival+hold} ($\lambda_h=0.5$) & 0.05 & 0.01 & 0.5 & 0.0242 & 0.0167 \\
\midrule
\multicolumn{6}{l}{\footnotesize $\dagger$ recomputed post-hoc; this run predates the arrival logging} \\
\bottomrule
\end{tabular}