Large Multimodal Model Evaluation on MLLM-as-a-Judge v1.0 (test)
49Overall ScoreGPT-4V
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPT-4Vsource=Reported in [6]2024.10 | 49 | 63.6 | 77.3 | |
| GPT-4osource=Author evaluation2024.10 | 43.9 | 57.7 | 73.6 | |
| GPT-4Vsource=Author evaluation2024.10 | 42.4 | 53.8 | 71.7 | |
| LLaVA-Critic-72B2024.10 | 39.3 | 57.8 | 71.5 | |
| LLaVA-Critic-7B2024.10 | 31.4 | 55.6 | 68.9 | |
| LLaVA-Critic-7BVersion=v0.52024.10 | 31.2 | 54.6 | 67.5 | |
| Gemini-prosource=Reported in [6]2024.10 | 30.4 | 50.9 | 61.5 | |
| LLaVA-OV-72B2024.10 | 28.7 | 51.3 | 70.1 | |
| LLaVA-CriticBackbone=LLaVA-NeXT2024.10 | 27.2 | 54.7 | 67.7 | |
| Qwen2-VL-7B-Instruct2024.10 | 25.3 | 34.8 | 64.5 | |
| LLaMA3.2-VBackbone=11B-Instruct2024.10 | 23.7 | 52.9 | 65.8 | |
| LLaVA-CriticBackbone=LLaVA-v1.5-7B2024.10 | 22.8 | 52.8 | 65.6 | |
| Prometheus-VBackbone=LLaVA-v1.5-7B2024.10 | 21.3 | — | — | |
| LLaVA-NeXTBackbone=LLaMA-8B2024.10 | 19.8 | 46.1 | 58.6 | |
| LLaVA-v1.5-7B2024.10 | 15.8 | 43.9 | 57.6 | |
| LLaVA-OV-7B2024.10 | 15.1 | 42.6 | 55 |