% Author: Simon-Pierre Boucher — simon-pierre.boucher@uqo.ca % generated by scripts/20_evaluate.py — do not edit \begin{tabular}{lrrrrrr} \toprule Arm & {Cost (USD)} & {Latency (s)} & {Tokens} & {Total cost (USD)} & {Mean $|e|$} & {N} \\ \midrule A (hedonic) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.219} & \num{1182} \\ B (hierarchical) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.207} & \num{1182} \\ C (catboost) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.215} & \num{1182} \\ C (gbm) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.218} & \num{1182} \\ C (xgboost) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.222} & \num{1182} \\ D (avm) & \num{0.0000} & \num{0.0} & \num{0} & \num{0.00} & \num{0.202} & \num{1182} \\ \midrule E (single llm) & \num{0.0205} & \num{16.1} & \num{6944} & \num{0.82} & \num{0.187} & \num{40} \\ F (independent) & \num{0.2341} & \num{54.7} & \num{75320} & \num{9.36} & \num{0.179} & \num{40} \\ G (debate r2) & \num{1.4996} & \num{244.9} & \num{514032} & \num{59.98} & \num{0.187} & \num{40} \\ \bottomrule \end{tabular}