Multi-image Hallucination Evaluation on BLINK
61.33AccuracyGLM4.1VBase + CAPL
Evaluation Results
| Method | Links | |
|---|---|---|
| GLM4.1VBase + CAPLParams=9B2026.03 | 61.33 | |
| GLM4.1VBaseParams=9B2026.03 | 58.17 | |
| Qwen2.5-VL + CAPLParams=7B2026.03 | 57.76 | |
| InternVL2.5 + CAPLParams=8B2026.03 | 55.76 | |
| InternVL2.5Params=8B2026.03 | 54.81 | |
| Qwen2.5-VLParams=7B2026.03 | 54.6 | |
| Qwen2VLParams=7B2026.03 | 53.17 | |
| InternVL2Params=7B2026.03 | 50.34 | |
| Idefics3Params=8B2026.03 | 48.34 | |
| Idefics2Params=8B2026.03 | 45.24 | |
| LLaVA-OVParams=7B2026.03 | 44.77 | |
| LLaVA-NextParams=7B2026.03 | 41.19 |