Short Video Question Answering on Vinoground
65.8Text Count/ScoreTarsier2-7B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Tarsier2-7BParameters=7B2025.01 | 65.8 | 38 | 28.8 | |
| GPT-4oModel Group=Proprietary2025.01 | 54 | 38.2 | 24.6 | |
| LLaVA-Video-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 52 | 35.6 | 20.8 | |
| Qwen2-VL-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 50.4 | 32.6 | 17.4 | |
| LLaVA-OV-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 48.4 | 35.2 | 21.8 | |
| LLaVA-OV-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 41.6 | 29.4 | 14.6 | |
| Previous SOTACitation=[52]2025.01 | 41.6 | 29.4 | 14.6 | |
| Qwen2-VL-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 40 | 23.4 | 12.4 | |
| Tarsier-34BModel Group=Open-source (>10B), Parameters=34B2025.01 | 37.8 | 32 | 15 | |
| LLaVA-Video-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 36.8 | 29 | 12.8 | |
| Gemini-1.5-ProModel Group=Proprietary2025.01 | 35.8 | 22.6 | 10.2 | |
| Tarsier-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 29.8 | 22.2 | 8.6 |