Action-relation hallucination evaluation on MMRel Real
78.56AccuracyRVE
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RVELVLM Model=InstructBLIP-7B2026.05 | 78.56 | 79.63 | |
| RVELVLM Model=LLaVA-1.5-13B2026.05 | 78.36 | 81.73 | |
| RVELVLM Model=LLaVA-1.5-7B2026.05 | 78.08 | 81.1 | |
| RVELVLM Model=LLaVA-NeXT-7B2026.05 | 77.39 | 81.23 | |
| RVELVLM Model=ShareGPT4V-7B2026.05 | 76.44 | 80.47 | |
| VanillaLVLM Model=LLaVA-1.5-13B2026.05 | 74.51 | 79.33 | |
| ICDLVLM Model=LLaVA-1.5-13B2026.05 | 74.06 | 79.01 | |
| ICDLVLM Model=LLaVA-1.5-7B2026.05 | 73.3 | 78.54 | |
| VCDLVLM Model=ShareGPT4V-7B2026.05 | 73.15 | 78.48 | |
| VanillaLVLM Model=ShareGPT4V-7B2026.05 | 72.96 | 78.44 | |
| VAFLVLM Model=LLaVA-1.5-13B2026.05 | 72.52 | 78.22 | |
| ICDLVLM Model=ShareGPT4V-7B2026.05 | 72.31 | 78.07 | |
| VCDLVLM Model=LLaVA-1.5-13B2026.05 | 71.8 | 77.77 | |
| VanillaLVLM Model=LLaVA-1.5-7B2026.05 | 71.12 | 77.32 | |
| ICDLVLM Model=LLaVA-NeXT-7B2026.05 | 70.79 | 77.34 | |
| VanillaLVLM Model=LLaVA-NeXT-7B2026.05 | 70.58 | 77.21 | |
| VAFLVLM Model=ShareGPT4V-7B2026.05 | 69.84 | 76.74 | |
| VAFLVLM Model=LLaVA-NeXT-7B2026.05 | 69.62 | 76.68 | |
| VCDLVLM Model=InstructBLIP-7B2026.05 | 68.97 | 79.62 | |
| VCDLVLM Model=LLaVA-NeXT-7B2026.05 | 68.86 | 76.3 | |
| VAFLVLM Model=LLaVA-1.5-7B2026.05 | 68.02 | 75.63 | |
| VanillaLVLM Model=InstructBLIP-7B2026.05 | 65.47 | 74.32 | |
| VCDLVLM Model=LLaVA-1.5-7B2026.05 | 64.67 | 73.83 | |
| VAFLVLM Model=InstructBLIP-7B2026.05 | 62.7 | 72.91 | |
| ICDLVLM Model=InstructBLIP-7B2026.05 | 61.1 | 71.99 |