                 initial model &                        dataset &            Llama-2-70b-chat-hf &     Mixtral-8x7B-Instruct-v0.1 &               Qwen1.5-72B-Chat  \\
                               &     \scalebox{0.9}[1]{MathGen} & \cellcolor[RGB]{240,230,140}{54.6} & \cellcolor[RGB]{240,230,140}{55.4} & \cellcolor[RGB]{240,230,140}{52.1} &                                 \\
                               &     \scalebox{0.9}[1]{FgFactV} & \cellcolor[RGB]{240,230,140}{53.8} & \cellcolor[RGB]{240,230,140}{49.3} & \cellcolor[RGB]{240,230,140}{47.5} &                                 \\
                               &      \scalebox{0.9}[1]{AnsCls} & \cellcolor[RGB]{240,230,140}{55.2} & \cellcolor[RGB]{240,230,140}{46.1} & \cellcolor[RGB]{240,230,140}{45.9} &                                 \\
                               &     \scalebox{0.9}[1]{MathGen} & \cellcolor[RGB]{240,230,140}{64.4} & \cellcolor[RGB]{240,230,140}{50.9} & \cellcolor[RGB]{240,230,140}{51.6} &                                 \\
                               &     \scalebox{0.9}[1]{FgFactV} & \cellcolor[RGB]{240,230,140}{70.2} & \cellcolor[RGB]{240,230,140}{39.1} & \cellcolor[RGB]{240,230,140}{29.7} &                                 \\
                               &      \scalebox{0.9}[1]{AnsCls} & \cellcolor[RGB]{240,230,140}{50.3} & \cellcolor[RGB]{240,230,140}{32.2} & \cellcolor[RGB]{240,230,140}{24.1} &                                 \\
