Time-related video understanding on E.T. Bench Subset
60.8REF ScoreMUSEG-7B
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MUSEG-7BCategory=Open-source ~ 7B Models2025.05 | 60.8 | 38.8 | 25.1 | 19 | 35.9 | |
| TEMPURACategory=Open-source ~ 3B Models2025.05 | 56.4 | 22.8 | 13.3 | 3.5 | 24 | |
| MUSEG-3BCategory=Open-source ~ 3B Models2025.05 | 54.3 | 28.7 | 18.3 | 11.8 | 28.3 | |
| Qwen2.5-VL-3BCategory=Open-source ~ 3B Models2025.05 | 52.9 | 20.4 | 12.7 | 7.6 | 23.4 | |
| Qwen2.5-VL-7BCategory=Open-source ~ 7B Models2025.05 | 51 | 30.3 | 16.5 | 9.3 | 26.8 | |
| Qwen2.5-VL-7B + vanilla GRPOCategory=Open-source ~ 7B Models, Training=vanilla GRPO2025.05 | 50.9 | 36.6 | 23.7 | 17.8 | 32.3 | |
| Video-R1Category=Open-source ~ 7B Models2025.05 | 49.2 | 22.2 | 15.6 | 12.8 | 25 | |
| VideoChat-R1Category=Open-source ~ 7B Models2025.05 | 47 | 35.9 | 24.1 | 12.5 | 29.9 | |
| TimeZeroCategory=Open-source ~ 7B Models2025.05 | 46.9 | 35.1 | 22.9 | 15.2 | 30 | |
| GPT-4oCategory=API-based Models2025.05 | 37.4 | 16.5 | 11.6 | 6.8 | 18.1 | |
| E.T. ChatCategory=Open-source ~ 7B Models2025.05 | 31.8 | 33.8 | 17.1 | 11.1 | 23.5 | |
| Qwen2.5-VL-7B + vanilla SFTCategory=Open-source ~ 7B Models, Training=vanilla SFT2025.05 | 27.8 | 12.6 | 15 | 8.7 | 16 | |
| TRACE-7BCategory=Open-source ~ 7B Models2025.05 | 25.2 | 17.2 | 14.7 | 5.3 | 15.6 |