Video Question Answering on SciVideoBench
29.8AccuracyTTA-Vid
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TTA-Vidbase_model=InternVL-3, #params=8B, #frames=322026.04 | 29.8 | 55.13 | |
| Video-R2#params=7B, #frames=1282026.04 | 28.4 | 53.92 | |
| Video-R1#params=7B, #frames=1282026.04 | 26.8 | 51.94 | |
| InternVL-3#params=8B, #frames=322026.04 | 26.5 | 51.37 | |
| VideoChat-R1#params=7B, #frames=1282026.04 | 26.5 | 52.34 | |
| VideoChat-R1.5#params=7B, #frames=1282026.04 | 25.8 | 52.24 | |
| Video-RFT#params=7B, #frames=1282026.04 | 25.7 | 52.32 | |
| TTA-Vidbase_model=Qwen2.5-VL, #params=7B, #frames=322026.04 | 25.7 | 51.84 | |
| TTRV*base_model=Qwen2.5-VL, #params=7B, #frames=322026.04 | 25.5 | 50.46 | |
| GPT-4o#params=−, #frames=−2026.04 | 24.9 | 60.02 | |
| LLaVA-OneVision#params=7B, #frames=642026.04 | 18.8 | 43.27 | |
| Qwen-2.5-VL#params=7B, #frames=<=7682026.04 | 16.4 | 48.82 | |
| Random#params=-, #frames=-2026.04 | 10 | 17.8 |