Multimodal Understanding on MM-Bench en (test)
83.9AccuracyPVC InternVL2 (Ours)
Evaluation Results
| Method | Links | |
|---|---|---|
| PVC InternVL2 (Ours)Size=8B, #token /image tile=642024.12 | 83.9 | |
| Qwen2-VLSize=7B, #token /image tile=Native resolution (unfixed)2024.12 | 83 | |
| IXC-2.5Size=7B, #token /image tile=4002024.12 | 82.2 | |
| InternVL2Size=8B, #token /image tile=2562024.12 | 81.7 | |
| Oryx MLLMSize=7B2024.12 | 81.4 | |
| LLaVA-NeXT-Yi-34BSize=34B2024.08 | 81.1 | |
| Cambrian-34BSize=34B2024.08 | 80.4 | |
| MiniCPM-Llama3-V 2.5Size=8.5B2024.08 | 77.2 | |
| GPT-4V (2023.11.06)2024.08 | 77 | |
| Bunny-Llama-3-8BSize=8.4B2024.08 | 77 | |
| PVC InternVL2 (Ours)Size=2B, #token /image tile=642024.12 | 75.8 | |
| Idefics2Size=8.0B2024.08 | 75.7 | |
| Qwen2-VLSize=2B, #token /image tile=Native resolution (unfixed)2024.12 | 74.9 | |
| Cambrian-8BSize=8.3B2024.08 | 74.6 | |
| CogVLM2-19B-ChatSize=19B2024.08 | 73.9 | |
| DeepSeek-VL-7BSize=7.3B2024.08 | 73.8 | |
| Gemini Pro2024.08 | 73.6 | |
| InternVL2Size=2B, #token /image tile=2562024.12 | 73.2 | |
| Yi-VL-34BSize=34B2024.08 | 72.4 | |
| XTuner-Llama-3-8B-v1.1Size=8.4B2024.08 | 71.7 | |
| MiniCPM-V 2.0Size=2.8B2024.08 | 69.1 | |
| Yi-VL-6BSize=6.7B2024.08 | 68.4 | |
| DeepSeek-VL-1.3BSize=1.7B2024.08 | 66.4 | |
| CogVLM-ChatSize=17B2024.08 | 65.8 | |
| MiniCPM-V 1.0Size=2.8B2024.08 | 64.1 | |
| Claude 3 Opus2024.08 | 63.3 | |
| Qwen-VL-ChatSize=9.6B2024.08 | 61.8 |