Real-world Understanding on WildVision
80.6Win RateGPT-4o-20240513
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-4o-20240513Model Name=GPT-4o-202405132024.12 | 80.6 | |
| GPT-4VModel Name=GPT-4V2024.12 | 71.8 | |
| InternVL2.5-78BModel Name=InternVL2.5-78B2024.12 | 71.4 | |
| LLaVA-OneVision-7B w/ UnifiedReward-ThinkReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 68.3 | |
| LLaVA-OneVision-7B w/ VL-MDRReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 68.3 | |
| LLaVA-OneVision-7B w/ Skywork-VLReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 68.2 | |
| LLaVA-OneVision-7B w/ IXC-2.5-RewardReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 67.9 | |
| LLaVA-OneVision-7B w/ UnifiedRewardReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 67.8 | |
| LLaVA-OneVision-7B w/ LLaVA-CriticReward Model Paradigm=Generative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 67.3 | |
| LLaVA-OneVision-7B w/ MM-RLHF-RewardReward Model Paradigm=Discriminative, Base Model=LLaVA-OneVision-7B, Alignment Method=DPO2026.04 | 67.1 | |
| InternVL2.5-38BModel Name=InternVL2.5-38B2024.12 | 66.4 | |
| InternVL2-Llama3-76BModel Name=InternVL2-Llama3-76B2024.12 | 65.8 | |
| InternVL2.5-26BModel Name=InternVL2.5-26B2024.12 | 65.2 | |
| InternVL2-40BModel Name=InternVL2-40B2024.12 | 63.2 | |
| InternVL2-26BModel Name=InternVL2-26B2024.12 | 62.2 | |
| InternVL2.5-8BModel Name=InternVL2.5-8B2024.12 | 62 | |
| InternVL-Chat-V1.5Model Name=InternVL-Chat-V1.52024.12 | 56.6 | |
| LLaVA-OneVision-7BBase Model=LLaVA-OneVision-7B2026.04 | 54.9 | |
| InternVL2-8BModel Name=InternVL2-8B2024.12 | 54.4 | |
| InternVL2.5-4BModel Name=InternVL2.5-4B2024.12 | 49.4 | |
| InternVL2.5-2BModel Name=InternVL2.5-2B2024.12 | 44.2 | |
| InternVL2-4BModel Name=InternVL2-4B2024.12 | 44.2 | |
| InternVL2.5-1BModel Name=InternVL2.5-1B2024.12 | 43.4 | |
| InternVL2-2BModel Name=InternVL2-2B2024.12 | 31.8 | |
| InternVL2-1BModel Name=InternVL2-1B2024.12 | 17.8 |