Multimodal Multilingual Evaluation on MMBench Multilingual
90Accuracy (en)InternVL2.5-78B
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| InternVL2.5-78BParameters=78B2024.12 | 90 | 89.7 | 87.4 | 83.3 | 84.9 | 86.3 | |
| InternVL2-Llama3-76BBackbone=Llama3, Parameters=76B2024.12 | 87.8 | 87.3 | 85.9 | 83.1 | 85 | 85.7 | |
| InternVL2.5-38BParameters=38B2024.12 | 87.5 | 88.6 | 85.3 | 84.5 | 84 | 85.9 | |
| Qwen2-VL-72BParameters=72B2024.12 | 86.9 | 87.2 | 85.8 | 83.5 | 84.4 | 85.3 | |
| InternVL2-40BParameters=40B2024.12 | 86.2 | 85.8 | 82.8 | 64 | 74.2 | 81.8 | |
| InternVL2.5-26BParameters=26B2024.12 | 86.1 | 85.5 | 80.7 | 67.5 | 75 | 79.6 | |
| InternVL2.5-8BParameters=8B2024.12 | 83.8 | 83.2 | 79.4 | 64.3 | 67.8 | 77.3 | |
| InternVL2-8BParameters=8B2024.12 | 82.9 | 81.8 | 76 | 60.5 | 66 | 74.4 | |
| InternVL2-26BParameters=26B2024.12 | 82.7 | 81.8 | 77.8 | 61.9 | 69.6 | 74.4 | |
| InternVL2.5-4BParameters=4B2024.12 | 82.3 | 81.1 | 78.9 | 73.4 | 68.1 | 76.2 | |
| Qwen2-VL-7BParameters=7B2024.12 | 81.8 | 81.6 | 79.1 | 75.6 | 74.5 | 79.3 | |
| InternVL-Chat-V1.5Version=V1.52024.12 | 81.1 | 80.2 | 76.9 | 56.2 | 66.7 | 71 | |
| GPT-4V2024.12 | 77.6 | 74.4 | 72.5 | 72.3 | 70.5 | 74.8 | |
| InternVL2-4BParameters=4B2024.12 | 77.3 | 72.4 | 72.6 | 43.6 | 46.5 | 61.2 | |
| InternVL2.5-2BParameters=2B2024.12 | 76.5 | 71.6 | 55.9 | 37.3 | 33.9 | 44.8 | |
| InternVL2-2BParameters=2B2024.12 | 73.8 | 69.6 | 51.4 | 29.8 | 31.3 | 42.3 | |
| Gemini-1.0-Pro2024.12 | 73.6 | 72.1 | 70.3 | 61.1 | 69.8 | 70.5 | |
| InternVL2.5-1BParameters=1B2024.12 | 72.5 | 64.7 | 57 | 43 | 37.8 | 53.2 | |
| Qwen2-VL-2BParameters=2B2024.12 | 72.1 | 71.1 | 69.9 | 61.1 | 54.4 | 69.3 | |
| InternVL2-1BParameters=1B2024.12 | 67.9 | 61.2 | 50.8 | 43.3 | 31.8 | 52.7 | |
| mPLUG-Owl22024.12 | 66.2 | 59.4 | 58.2 | 37.9 | 47.7 | 60.4 |