Image Question Answering on MMBench (MMB-EN/CN)
81MMB-EN ScoreLLaVA-OV-7B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLaVA-OV-7BKV Size=1962026.05 | 81 | 82.2 | |
| LLaVA-OV-7B + HERMESKV Size=1002026.05 | 79.8 | 80.6 | |
| Qwen2.5-VL-3BKV Size=2562026.05 | 77 | 77.4 | |
| Cambrian-4B + Gaze attentionKV Size=288+42026.05 | 76.4 | 74.8 | |
| Cambrian-4B + Gaze attentionKV Size=144+42026.05 | 76.3 | 76.3 | |
| LLaVA-Next-Qwen2-7BKV Size=7842026.05 | 75.2 | — | |
| LLaVA-OV-7B + HERMESKV Size=402026.05 | 75.2 | 74.8 | |
| Cambrian-4BKV Size=5762026.05 | 74.9 | 73.4 | |
| Qwen2.5-VL-3B + InfiniPot-VKV Size=1282026.05 | 74.7 | 74.1 | |
| Cambrian-4B + TokenCompre. + Gaze attentionKV Size=54+42026.05 | 74.7 | 71.6 | |
| Cambrian-4B + Gaze attentionKV Size=72+42026.05 | 74.6 | 75 | |
| Cambrian-4B + TokenCompre.KV Size=1442026.05 | 74.3 | 73 | |
| LLaVA-Next-LLaMA-3-8BKV Size=7842026.05 | 74 | — | |
| Cambrian-4B + TokenCompre. + Gaze attentionKV Size=36+42026.05 | 74 | 73.1 | |
| Cambrian-8B-CLIP-LKV Size=5762026.05 | 72.7 | — | |
| Qwen2.5-VL-3B + InfiniPot-VKV Size=502026.05 | 68.5 | 67.6 | |
| Cambrian-8B-DINOv2-LKV Size=5762026.05 | 65.6 | — |