Multimodal Reasoning on MMBench EN
83AccuracyInstruct
Evaluation Results
| Method | Links | |
|---|---|---|
| InstructModel Scale=14B, Base Model=InternVL32026.03 | 83 | |
| DPOModel Scale=14B, Base Model=InternVL32026.03 | 83 | |
| DPOModel Scale=8B, Base Model=InternVL32026.03 | 83 | |
| DPO-ShiftModel Scale=14B, Base Model=InternVL32026.03 | 83 | |
| ACPOModel Scale=14B, Base Model=InternVL32026.03 | 83 | |
| InstructModel Scale=8B, Base Model=InternVL32026.03 | 82.8 | |
| IPOModel Scale=14B, Base Model=InternVL32026.03 | 82 | |
| IPOModel Scale=8B, Base Model=InternVL32026.03 | 82 | |
| β-DPOModel Scale=14B, Base Model=InternVL32026.03 | 82 | |
| SimPOModel Scale=14B, Base Model=InternVL32026.03 | 81.5 | |
| SimPOModel Scale=8B, Base Model=InternVL32026.03 | 81 | |
| β-DPOModel Scale=8B, Base Model=InternVL32026.03 | 81 | |
| DPO-ShiftModel Scale=8B, Base Model=InternVL32026.03 | 81 | |
| ACPOModel Scale=8B, Base Model=InternVL32026.03 | 81 | |
| F3ARatio=40%2026.05 | 77.73 | |
| Qwen3-VL-2BRatio=100%2026.05 | 77.5 | |
| DivPruneRatio=60%2026.05 | 77.5 | |
| VisionZipRatio=60%2026.05 | 77.5 | |
| F3ARatio=60%2026.05 | 77.5 | |
| CDPrunerRatio=60%2026.05 | 77.08 | |
| CDPrunerRatio=40%2026.05 | 77.02 | |
| DivPruneRatio=40%2026.05 | 76.83 | |
| VisionZipRatio=40%2026.05 | 76.46 | |
| F3ARatio=20%2026.05 | 76.07 | |
| FastVRatio=60%2026.05 | 76 | |
| CDPrunerRatio=20%2026.05 | 75.93 | |
| FastVRatio=40%2026.05 | 74.66 | |
| VisionZipRatio=20%2026.05 | 74.5 | |
| DivPruneRatio=20%2026.05 | 74.47 | |
| FastVRatio=20%2026.05 | 72.07 | |
| LLaVA-1.5-13B + MMFuserLLM=Vicuna-13B2024.10 | 69.9 | |
| LLaVA-1.5-13BLLM=Vicuna-13B2024.10 | 67.7 | |
| LLaVA-1.5-7B + MMFuserLLM=Vicuna-7B2024.10 | 67.5 | |
| LLaVA-1.5-7BLLM=Vicuna-7B2024.10 | 64.3 | |
| Qwen-VL-ChatLLM=Qwen-7B2024.10 | 60.6 | |
| ShikraLLM=Vicuna-13B2024.10 | 58.8 | |
| IDEFICS-80BLLM=LLaMA-65B2024.10 | 54.5 | |
| IDEFICS-9BLLM=LLaMA-7B2024.10 | 48.2 | |
| Qwen-VLLLM=Qwen-7B2024.10 | 38.2 | |
| InstructBLIPLLM=Vicuna-7B2024.10 | 36 |