% SFT Data Statistics
% Statistics of the training data for SFT experiments
\begin{table}[t]
\centering
\caption{Statistics of SFT training data. This table summarizes the revision-format training split only; it is kept in the appendix because the main text now focuses on the benchmark and evaluation tables.}
\label{app:sft_data_statistics}
\resizebox{\columnwidth}{!}{
\begin{tabular}{lcccc}
\toprule
\textbf{Language} & \textbf{\#Samples} & \textbf{Avg. Messages} & \textbf{Rule Types} & \textbf{Linter} \\
\midrule
C++ & 538 & 10.4 & 89 & clang-tidy \\
Python & 416 & 8.9 & 72 & Pylint \\
Java & 242 & 18.8 & 63 & PMD \\
\midrule
Total & 1,196 & 11.6 & 224 & -- \\
\bottomrule
\end{tabular}
}
\end{table}