Multi-modal Understanding on MM-Vet v1 (full)
36.2Overall Score (MM-Vet v1)Ours
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| OursModel Size=13B2025.07 | 36.2 | 38.9 | 30.2 | 22.6 | 23.1 | 32.7 | 15 | |
| LLaVA-v1.5-13BModel Size=13B2025.07 | 36 | 39.7 | 28.8 | 23.2 | 24.2 | 34.5 | 11.5 | |
| vanilla-DPOModel Size=13B2025.07 | 35 | 38.4 | 29.7 | 17.9 | 21 | 35.6 | 11.5 | |
| VCDModel Size=13B2025.07 | 33.7 | 38.7 | 24.4 | 22.4 | 26.4 | 30.1 | 7.7 | |
| HSA-DPOModel Size=13B2025.07 | 33.7 | 35.9 | 28.4 | 16.4 | 18.9 | 34.5 | 15 | |
| Ours + HA-DPOModel Size=7B2025.07 | 33.5 | 38.4 | 25 | 21.2 | 23.7 | 29.3 | 7.7 | |
| OursModel Size=7B2025.07 | 32.6 | 37.7 | 23.1 | 22.7 | 25.6 | 26.8 | 7.7 | |
| POVIDModel Size=7B2025.07 | 31.8 | — | — | — | — | — | — | |
| EFUFModel Size=7B2025.07 | 31.2 | 36.5 | 21.4 | 17.1 | 19.5 | 27.9 | 7.7 | |
| LLaVA-v1.5-7BModel Size=7B2025.07 | 31 | 35.9 | 23.3 | 17.1 | 22 | 25.9 | 11.5 | |
| DoLaModel Size=7B2025.07 | 30.8 | 36.1 | 21.3 | 19.4 | 20.9 | 26.9 | 7.7 | |
| HA-DPOModel Size=7B2025.07 | 30.6 | 35.5 | 22.1 | 18.3 | 21.9 | 26.3 | 7.7 | |
| OPERAModel Size=7B2025.07 | 30.3 | 34.9 | 21.6 | 18.7 | 21.1 | 25.7 | 7.7 | |
| RLAIF-VModel Size=7B2025.07 | 29.9 | 34.4 | 23.4 | 18.7 | 23.7 | 27.7 | 7.3 | |
| VCDModel Size=7B2025.07 | 29.8 | 34.5 | 21.9 | 18.3 | 20.6 | 24.8 | 3.8 | |
| TPOModel Size=7B2025.07 | 25.7 | 31.8 | 15.4 | 16.7 | 19.6 | 22.1 | 7.7 |