Image Captioning on CHAIR
69.8CHAIR_SVISTA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| VISTAModel=mPLUG-Owl2-7B2026.06 | 69.8 | 42.1 | 78.7 | 105.5 | |
| DAMOModel=InstructBLIP-7B2026.06 | 65.6 | 39.5 | 73.5 | 104.7 | |
| VCDModel=mPLUG-Owl2-7B2026.06 | 64.4 | 18.1 | 77.9 | 110.4 | |
| GreedyModel=InstructBLIP-7B2026.06 | 63.4 | 38.5 | 73.7 | 101.6 | |
| VCDModel=InstructBLIP-7B2026.04 | 59 | 14.8 | 72 | — | |
| VCDBackbone=Shikra2026.06 | 59 | 15.58 | — | — | |
| FPSPrunerPruning strategy category=Diversity-based2026.03 | 58.6 | 18.6 | 76 | 100.5 | |
| VCDModel=LLaVA-1.5-7B2026.06 | 58.6 | 16.5 | 82.1 | 105.5 | |
| DAMOModel=mPLUG-Owl2-7B2026.06 | 58.6 | 17.5 | 77.3 | 106.4 | |
| regularModel=InstructBLIP-7B2026.04 | 58 | 16.3 | 71.1 | — | |
| GreedyModel=mPLUG-Owl2-7B2026.06 | 57.8 | 17.1 | 78.6 | 105.6 | |
| SamplingBackbone=LLAVA-1.52026.06 | 57.6 | 17.18 | — | — | |
| ICDBackbone=Shikra2026.06 | 57.6 | 14.68 | — | — | |
| DivPrunePruning strategy category=Diversity-based2026.03 | 57.4 | 18 | 76.4 | 101.1 | |
| PAIModel=mPLUG-Owl2-7B2026.06 | 57.4 | 14.5 | 79.7 | 105.7 | |
| VCDModel=InstructBLIP-7B2026.06 | 57.2 | 40.1 | 63.5 | 87.8 | |
| DAMOModel=LLaVA-1.5-7B2026.06 | 56.6 | 16.7 | 81.6 | 106.7 | |
| SamplingBackbone=Shikra2026.06 | 56.4 | 15.96 | — | — | |
| VCDBackbone=InstructBLIP2026.06 | 56.2 | 16.65 | — | — | |
| VAFModel=InstructBLIP-7B2026.04 | 56 | 15.1 | 70.4 | — | |
| SamplingBackbone=InstructBLIP2026.06 | 56 | 16.91 | — | — | |
| ICDBackbone=InstructBLIP2026.06 | 55.4 | 16.52 | — | — | |
| FADEModel=mPLUG-Owl2-7B2026.06 | 55 | 16.6 | 76.3 | 105.4 | |
| SIDBackbone=Shikra2026.06 | 54.2 | 14.79 | — | — | |
| VCDBackbone=LLAVA-1.52026.06 | 54 | 16.02 | — | — | |
| SIDBackbone=InstructBLIP2026.06 | 54 | 16.89 | — | — | |
| VISTAModel=InstructBLIP-7B2026.06 | 54 | 41.3 | 60.4 | 85.6 | |
| OPERABackbone=Shikra2026.06 | 53.6 | 14.2 | — | — | |
| ICDBackbone=LLAVA-1.52026.06 | 53.4 | 15.67 | — | — | |
| SIDBackbone=LLAVA-1.52026.06 | 53.4 | 15.4 | — | — | |
| VAFModel=LLaVA1.5-7B2026.04 | 53 | 16.5 | 76.9 | — | |
| CausalMMBackbone=Shikra2026.06 | 53 | 14.27 | — | — | |
| FoxBackbone=Shikra2026.06 | 52 | 13.66 | — | — | |
| LLaVA-1.5-7B2026.03 | 51 | 13.9 | 78.7 | 101.4 | |
| regularModel=LLaVA1.5-7B2026.04 | 51 | 17.6 | 74.4 | — | |
| VCDModel=LLaVA1.5-7B2026.04 | 51 | 16.7 | 77.2 | — | |
| VisprunerPruning strategy category=Attention-based2026.03 | 49.8 | 15 | 72.6 | 96.7 | |
| GreedyModel=LLaVA-1.5-7B2026.06 | 49.8 | 14.8 | 80.6 | 101.2 | |
| FADEModel=InstructBLIP-7B2026.06 | 49.2 | 14 | 72.9 | 99.8 | |
| OPERABackbone=LLAVA-1.52026.06 | 49 | 13.52 | — | — | |
| PAIModel=InstructBLIP-7B2026.06 | 48.6 | 37.9 | 58.1 | 68.8 | |
| CausalMMBackbone=LLAVA-1.52026.06 | 47.6 | 13.49 | — | — | |
| AGLAModel=LLaVA1.5-7B2026.04 | 47 | 14.2 | 77.8 | — | |
| FADEModel=LLaVA-1.5-7B2026.06 | 46.6 | 14.1 | 78.7 | 98.6 | |
| FoxBackbone=LLAVA-1.52026.06 | 46.4 | 12.9 | — | — | |
| PNDModel=LLaVA1.5-7B2026.04 | 46 | 14 | 78.1 | — | |
| AGLAModel=InstructBLIP-7B2026.04 | 46 | 12.3 | 71.5 | — | |
| FasterVLMPruning strategy category=Attention-based2026.03 | 45.4 | 13.5 | 69.3 | 94 | |
| PruMerge+Pruning strategy category=Attention-based2026.03 | 45.2 | 15.6 | 66.7 | 91.4 | |
| OPERABackbone=InstructBLIP2026.06 | 45 | 13.15 | — | — | |
| CausalMMBackbone=InstructBLIP2026.06 | 43.2 | 13.05 | — | — | |
| FoxBackbone=InstructBLIP2026.06 | 42.4 | 11.98 | — | — | |
| PNDModel=InstructBLIP-7B2026.04 | 42 | 11.2 | 72.1 | — | |
| PAIModel=LLaVA-1.5-7B2026.06 | 35.6 | 9.8 | 74.8 | 107.6 | |
| RegularBase Model=InstructBLIP2025.12 | 31.3 | 11 | 59.1 | 53.5 | |
| WoodpeckerBase Model=InstructBLIP2025.12 | 31.1 | 10.7 | 62.4 | 51.4 | |
| M3IDBase Model=InstructBLIP2025.12 | 30.7 | 10.3 | 62.5 | 53.3 | |
| VCDBase Model=InstructBLIP2025.12 | 29.9 | 10.2 | 61.9 | 54.1 | |
| RITUALBase Model=InstructBLIP2025.12 | 26.7 | 8.8 | 63.5 | 55.2 | |
| RegularBase Model=LLaVA-1.52025.12 | 26.1 | 9.3 | 58.6 | 53.3 | |
| WoodpeckerBase Model=LLaVA-1.52025.12 | 24.8 | 7.6 | 60.9 | 49.6 | |
| HALCBase Model=InstructBLIP2025.12 | 24.6 | 8.1 | 64.2 | 54.8 | |
| VCDBase Model=LLaVA-1.52025.12 | 24.3 | 7.8 | 63.4 | 54.4 | |
| DeFGBase Model=InstructBLIP2025.12 | 24 | 7.7 | 63.5 | 55 | |
| CoFi-DecBase Model=InstructBLIP2025.12 | 23.2 | 7.3 | 68.9 | 55.8 | |
| RITUALBase Model=LLaVA-1.52025.12 | 22.3 | 6.8 | 63.1 | 54.7 | |
| HALCBase Model=LLaVA-1.52025.12 | 21.6 | 7.2 | 64.9 | 53.5 | |
| M3IDBase Model=LLaVA-1.52025.12 | 21.5 | 6.1 | 64.1 | 53.6 | |
| VISTAModel=LLaVA-1.5-7B2026.06 | 19.2 | 6.5 | 62.6 | 86.2 | |
| DeFGBase Model=LLaVA-1.52025.12 | 18.4 | 6.4 | 62.9 | 54.3 | |
| CoFi-DecBase Model=LLaVA-1.52025.12 | 18.1 | 6.1 | 65.2 | 55.7 |