Medical Visual Question Answering (True/False) on IU-XRay (test)
97.78AccuracyLIAR
Evaluation Results
| Method | Links | |
|---|---|---|
| LIARModel=GPT-5, Retriever=Filtered2026.05 | 97.78 | |
| LIARModel=GPT-5, Retriever=BGE-VL2026.05 | 95.78 | |
| LIARModel=GPT-5, Retriever=CLIP2026.05 | 94.6 | |
| Clean RAG (avg.)Model=GPT-5, Retriever=Average2026.05 | 93.9 | |
| M³AttModel=GPT-5, Retriever=BGE-VL2026.05 | 93.54 | |
| M³AttModel=GPT-5, Retriever=CLIP2026.05 | 91.96 | |
| Clean RAG (avg.)Model=GPT-4o, Retriever=Average2026.05 | 89.64 | |
| w/o RAGModel=GPT-5, Retriever=None2026.05 | 88.28 | |
| M³AttModel=GPT-5, Retriever=Filtered2026.05 | 87.44 | |
| LIARModel=GPT-4o, Retriever=BGE-VL2026.05 | 86.04 | |
| LIARModel=GPT-4o, Retriever=CLIP2026.05 | 83.9 | |
| M³AttModel=GPT-4o, Retriever=BGE-VL2026.05 | 80.44 | |
| LIARModel=Gemini-2.5, Retriever=CLIP2026.05 | 80.02 | |
| M³AttModel=GPT-4o, Retriever=CLIP2026.05 | 77.88 | |
| LIARModel=LLaVA-Med, Retriever=CLIP2026.05 | 76.92 | |
| M³AttModel=Gemini-2.5, Retriever=CLIP2026.05 | 76.12 | |
| LIARModel=Gemini-2.5, Retriever=Filtered2026.05 | 74.88 | |
| Clean RAG (avg.)Model=Gemini-2.5, Retriever=Average2026.05 | 72.6 | |
| Clean RAG (avg.)Model=LLaVA-Med, Retriever=Average2026.05 | 72.51 | |
| LIARModel=GPT-4o, Retriever=Filtered2026.05 | 71.06 | |
| w/o RAGModel=Gemini-2.5, Retriever=None2026.05 | 68.2 | |
| w/o RAGModel=GPT-4o, Retriever=None2026.05 | 67.36 | |
| M³AttModel=Gemini-2.5, Retriever=Filtered2026.05 | 66.52 | |
| M³AttModel=LLaVA-Med, Retriever=CLIP2026.05 | 65.28 | |
| M³AttModel=GPT-4o, Retriever=Filtered2026.05 | 61.5 | |
| M³AttModel=Gemini-2.5, Retriever=BGE-VL2026.05 | 59.84 | |
| LIARModel=Gemini-2.5, Retriever=BGE-VL2026.05 | 58.78 | |
| LIARModel=Claude-4.5, Retriever=CLIP2026.05 | 55.9 | |
| Clean RAG (avg.)Model=Claude-4.5, Retriever=Average2026.05 | 53.2 | |
| LIARModel=LLaVA-Med, Retriever=BGE-VL2026.05 | 51.96 | |
| M³AttModel=Claude-4.5, Retriever=CLIP2026.05 | 47.04 | |
| M³AttModel=LLaVA-Med, Retriever=BGE-VL2026.05 | 46.56 | |
| LIARModel=Claude-4.5, Retriever=BGE-VL2026.05 | 43.26 | |
| M³AttModel=Claude-4.5, Retriever=BGE-VL2026.05 | 38.3 | |
| LIARModel=LLaVA-Med, Retriever=Filtered2026.05 | 33.46 | |
| LIARModel=Claude-4.5, Retriever=Filtered2026.05 | 28.84 | |
| M³AttModel=LLaVA-Med, Retriever=Filtered2026.05 | 28.04 | |
| M³AttModel=Claude-4.5, Retriever=Filtered2026.05 | 20.5 | |
| w/o RAGModel=LLaVA-Med, Retriever=None2026.05 | 14.64 | |
| w/o RAGModel=Claude-4.5, Retriever=None2026.05 | 0 |