Visual Question Answering on AnimalKB
87.99Accuracy (Visual)InternVL2.5-26B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| InternVL2.5-26BSize=25.5 B, KB Usage=with KB2025.12 | 87.99 | 88.04 | 88.01 | |
| Qwen2-VL-7BSize=8.3 B, KB Usage=with KB2025.12 | 87.67 | 86.91 | 87.29 | |
| InternVL2.5-8BSize=8.1 B, KB Usage=with KB2025.12 | 87.23 | 86.47 | 86.85 | |
| Qwen2.5-VL-7BSize=8.3 B, KB Usage=with KB2025.12 | 87.06 | 85.42 | 86.24 | |
| DeepSeek-VL2-SmallSize=16.1 B, KB Usage=with KB2025.12 | 85.39 | 84.73 | 85.06 | |
| DeepSeek-VL2Size=27.5 B, KB Usage=with KB2025.12 | 84.66 | 85.1 | 84.88 | |
| LLaVA-Next-Llama3-8BSize=8.4 B, KB Usage=with KB2025.12 | 83.41 | 84.19 | 83.8 | |
| LLaVA-OV-Qwen2-7BSize=8.0 B, KB Usage=with KB2025.12 | 83.28 | 81.81 | 82.55 | |
| LLaVA-v1.6-Vicuna-7BSize=7.1 B, KB Usage=with KB2025.12 | 80.71 | 80.86 | 80.78 | |
| InternVL2.5-26BSize=25.5 B, KB Usage=without KB2025.12 | 76.52 | 72.55 | 74.53 | |
| Qwen2-VL-7BSize=8.3 B, KB Usage=without KB2025.12 | 74.56 | 72.84 | 73.7 | |
| Qwen2.5-VL-7BSize=8.3 B, KB Usage=without KB2025.12 | 74.44 | 71.25 | 72.84 | |
| DeepSeek-VL2Size=27.5 B, KB Usage=without KB2025.12 | 73.75 | 75.69 | 74.72 | |
| LLaVA-OV-Qwen2-7BSize=8.0 B, KB Usage=without KB2025.12 | 71.94 | 65.27 | 68.6 | |
| InternVL2.5-8BSize=8.1 B, KB Usage=without KB2025.12 | 71.27 | 69.09 | 70.18 | |
| DeepSeek-VL2-SmallSize=16.1 B, KB Usage=without KB2025.12 | 66.13 | 68.7 | 67.41 | |
| LLaVA-v1.6-Vicuna-7BSize=7.1 B, KB Usage=without KB2025.12 | 65.76 | 60.91 | 63.33 | |
| LLaVA-Next-Llama3-8BSize=8.4 B, KB Usage=without KB2025.12 | 65.47 | 63.75 | 64.61 |