General Multimodal Evaluation on Macro-average of HallusionBench, AMBER, CRPE, R-Bench, and BLINK
63.35Overall ScoreIC-VCO
Evaluation results
| Method | Links | |
|---|---|---|
| IC-VCOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 63.35 | |
| IC-VCOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 62.83 | |
| SymMPOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 62.11 | |
| mDPOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 62.02 | |
| mDPOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 61.64 | |
| SymMPOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 61.5 | |
| S-VCOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 61.41 | |
| S-VCOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 60.81 | |
| DPOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 60.4 | |
| V-DPOContrastive Sample Source=Contrastive Sample Editing, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 60.38 | |
| DPOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 60.32 | |
| V-DPOContrastive Sample Source=Synthetic, Backbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 60.15 | |
| LLaVA-NeXT-Interleave-Qwen-7BBackbone=LLaVA-NeXT-Interleave-Qwen-7B2026.05 | 59.14 |