Question Answering on MLLMU-Bench Forget 1.0
55AccuracyVanilla
Evaluation results
| Method | Links | |
|---|---|---|
| VanillaBackbone=Qwen2-VL-7B2025.12 | 55 | |
| Visual Knowledge DistillationBackbone=Qwen2-VL-7B2025.12 | 54.6 | |
| MMUBackbone=Qwen2-VL-7B2025.12 | 54.4 | |
| GA DiffBackbone=Qwen2-VL-7B2025.12 | 52.8 | |
| NPOBackbone=Qwen2-VL-7B2025.12 | 50.4 | |
| MANUBackbone=Qwen2-VL-7B2025.12 | 47.2 | |
| GABackbone=Qwen2-VL-7B2025.12 | 46.7 | |
| KL MinBackbone=Qwen2-VL-7B2025.12 | 45.3 | |
| VanillaBackbone=LLaVA-1.5-7B2025.12 | 38.4 | |
| Visual Knowledge DistillationBackbone=LLaVA-1.5-7B2025.12 | 35.8 | |
| MMUBackbone=LLaVA-1.5-7B2025.12 | 34.2 | |
| GA DiffBackbone=LLaVA-1.5-7B2025.12 | 33.6 | |
| KL MinBackbone=LLaVA-1.5-7B2025.12 | 33.6 | |
| NPOBackbone=LLaVA-1.5-7B2025.12 | 33.6 | |
| MANUBackbone=LLaVA-1.5-7B2025.12 | 33.6 | |
| GABackbone=LLaVA-1.5-7B2025.12 | 32.5 |