Visual Question Answering on GQA (Score/Gain)
63.3ScoreFast-dVLM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Fast-dVLMModel Category=Diffusion, Decoding Strategy=spec.2026.04 | 63.3 | — | |
| Fast-dVLMModel Category=Diffusion, Decoding Strategy=MDM2026.04 | 63 | — | |
| LLaVA(base)2025.03 | 61.94 | — | |
| AdaMMSUnsupervised=true2025.03 | 61.81 | 2.79 | |
| VILA-1.5-3BModel Category=Autoregressive2026.04 | 61.4 | — | |
| Intern-VL-2.5-4BModel Category=Autoregressive2026.04 | 61 | — | |
| Task ArithmeticUnsupervised=false2025.03 | 59.68 | 0.66 | |
| DARE-LinearUnsupervised=false2025.03 | 59.56 | 0.54 | |
| DimpleModel Category=Diffusion2026.04 | 59.2 | — | |
| Qwen2.5-VL-3BModel Category=Autoregressive2026.04 | 59 | — | |
| Ties-MergingUnsupervised=false2025.03 | 58.37 | -0.65 | |
| MetaGPTUnsupervised=true2025.03 | 58.37 | -0.66 | |
| DARE-TiesUnsupervised=false2025.03 | 57.51 | -1.51 | |
| mPLUG-Owl22025.03 | 56.11 | — | |
| LaViDaModel Category=Diffusion2026.04 | 55.5 | — | |
| LLaDA-VModel Category=Diffusion2026.04 | 53.4 | — | |
| MiniCPM-V-2 (3B)Model Category=Autoregressive2026.04 | 51.7 | — |