Short Video Question Answering on TVBench (test)
54.7AccuracyTarsier2-7B
Evaluation results
| Method | Links | |
|---|---|---|
| Tarsier2-7BParameters=7B2025.01 | 54.7 | |
| Tarsier-34BModel Group=Open-source (>10B), Parameters=34B2025.01 | 53.8 | |
| Qwen2-VL-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 52.7 | |
| Previous SOTACitation=[124]2025.01 | 51.6 | |
| LLaVA-Video-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 50 | |
| Gemini-1.5-ProModel Group=Proprietary2025.01 | 46.5 | |
| LLaVA-OV-72BModel Group=Open-source (>10B), Parameters=72B2025.01 | 45.9 | |
| Tarsier-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 45.8 | |
| LLaVA-OV-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 45.6 | |
| LLaVA-Video-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 45.6 | |
| Qwen2-VL-7BModel Group=Open-source (<=10B), Parameters=7B2025.01 | 43.8 | |
| GPT-4oModel Group=Proprietary2025.01 | 39.6 |