\begin{table}[t] \centering \caption{Main result.} \resizebox{1.0\linewidth}{!}{% \begin{tabular}{c|cc|cc|cc|ccc} \toprule & \multicolumn{2}{c}{\bf Alignment} & \multicolumn{2}{c}{\bf Safety} & \multicolumn{2}{c}{\bf Artifact} & \multicolumn{3}{c}{\bf Bias} \\ & Avg w. tie & Avg w.o. Tie & Avg w. tie & Avg w.o. Tie & Avg w. tie & Avg w.o. Tie & ACC & NDS & GES \\ \midrule CLIP-v1$^\diamondsuit$ & $44.0$ & $60.7$ & $13.1$ & $25.7$ & $41.9$ & $82.7$ & 57.4 & 76.3 & 86.9 \\ BLIP-v2$^\diamondsuit$ & $21.5$ & $34.1$ & $44.3$ & $75.3$ & $7.8$ & $24.4$ & 68.7 & 83.7 & 91.3 \\ PickScore-v1$^\diamondsuit$ & $60.9$ & $65.9$ & $37.3$ & $41.3$ & $83.9$ & $92.2$ & 31.0 & 66.5 & 81.1 \\ HPS-v2.1$^\diamondsuit$ & $48.8$ & $73.6$ & $20.8$ & $35.7$ & $69.6$ & $99.1$ & 55.0 & 77.9 & 87.6 \\ ImageReward$^\diamondsuit$ & $51.1$ & $67.9$ & $24.9$ & $35.9$ & $63.5$ & $91.7$ & 40.9 & 73.7 & 85.3 \\ Aesthetics$^\diamondsuit$ & $34.8$ & $56.7$ & $31.6$ & $54.7$ & $70.8$ & $98.5$ & 61.4 & 85.7 & 92.1 \\ \midrule LLaVA-1.5-7b$^\heartsuit$ & $22.0$ & $50.8$ & - & - & - & - & 83.7 & 70.4 & 88.7 \\ LLaVA-1.5-13b$^\heartsuit$ & $10.3$ & $51.9$ & - & - & - & - & 69.7 & 74.3 & 88.6 \\ LLaVA-NeXT-mistral-7b$^\heartsuit$ & - & - & - & - & - & - & 69.9 & 64.3 & 85.4 \\ LLaVA-NeXT-vicuna-13b$^\heartsuit$ & - & - & - & - & - & - & 56.3 & 64.0 & 82.7 \\ Instructblip-7b$^\heartsuit$ & - & - & - & - & - & - & 53.1 & 80.8 & 91.2 \\ MiniGPT4-v2$^\heartsuit$ & - & - & - & - & - & - & 32.6 & 67.0 & 83.3 \\ Prometheus-Vision-7b$^\heartsuit$ & - & - & - & - & - & - & 49.5 & 43.4 & 74.4 \\ Prometheus-Vision-13b$^\heartsuit$ & - & - & - & - & - & - & 66.3 & 46.3 & 76.8 \\ Qwen-VL-Chat$^\heartsuit$ & $31.1$ & $31.6$ & - & - & - & - & 71.9 & 62.8 & 86.2 \\ Internvl-chat-v1-5$^\heartsuit$ & $75.8$ & $77.6$ & - & - & - & - & 25.4 & 69.6 & 84.3 \\ Idefics2-8b$^\heartsuit$ & $32.6$ & $43.5$ & - & - & - & - & 42.1 & 58.7 & 79.4 \\ \midrule Qwen-VL-Chat$^\spadesuit$ & $31.1$ & $31.6$ & - & - & - & - & 71.9 & 62.8 & 86.2 \\ Internvl-chat-v1-5$^\spadesuit$ & $75.8$ & $77.6$ & - & - & - & - & 25.4 & 69.6 & 84.3 \\ Idefics2-8b$^\spadesuit$ & $32.6$ & $43.5$ & - & - & - & - & 42.1 & 58.7 & 79.4 \\ GPT-4-vision$^\clubsuit$ & - & - & - & - & - & - & 79.0 & 80.4 & 93.2 \\ GPT-4o$^\clubsuit$ & - & - & - & - & - & - & 65.8 & 82.5 & 92.8 \\ Gemini Ultra$^\clubsuit$ & - & - & - & - & - & - & 55.6 & 75.3 & 88.6 \\ Claude 3 Opus$^\clubsuit$ & - & - & - & - & - & - & 57.7 & 65.6 & 85.0 \\ \bottomrule \end{tabular}% } % \label{exp:main_result} \end{table}