Visual Multiple Choice Question Answering on MC-LLaVA
90.9Choice-V Accuracy (Single)MC-LLaVA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MC-LLaVAT (Tokens)=10^1, Auxiliary Loss=true2024.11 | 90.9 | 92 | 91.4 | |
| GPT4o+PromptT (Tokens)=10^12024.11 | 88.8 | 88.9 | 88.9 | |
| MC-LLaVAT (Tokens)=10^1, Auxiliary Loss=false2024.11 | 87.7 | 90.5 | 89 | |
| PeKitT (Tokens)=02024.11 | 87.4 | 86.7 | 87.2 | |
| LLaVA+PromptT (Tokens)=10^22024.11 | 84.1 | 78.5 | 82.5 | |
| LLaVA+PromptT (Tokens)=10^12024.11 | 83.7 | 78.1 | 81.7 | |
| RAP-MLLMT (Tokens)=10^22024.11 | 83.2 | 69 | 78.4 | |
| LLaVAT (Tokens)=02024.11 | 80.6 | 80.2 | 80.4 | |
| Yo'LLaVA-ST (Tokens)=10^12024.11 | 80.1 | — | 80.1 | |
| MyVLMT (Tokens)=12024.11 | 77.9 | — | 77.9 | |
| Yo'LLaVA-MT (Tokens)=10^12024.11 | 68.8 | 60.2 | 65.5 |