Grounded Video Question Answering on ReXTime (test)
43.2mIoUZoom-Zero
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Zoom-ZeroModel Scale=7B-8B, Training Paradigm=RL-based2025.12 | 43.2 | 62 | 56.5 | 44.1 | |
| VideoChat-R1Model Scale=7B-8B, Training Paradigm=RL-based2025.12 | 38.6 | 58.1 | 50.6 | 39 | |
| TVG-R1Model Scale=7B-8B, Training Paradigm=RL-based2025.12 | 28.2 | 53.6 | 41 | 24.5 | |
| Qwen2.5-VLModel Scale=7B-8B2025.12 | 27.4 | 51.1 | 36.1 | 24.8 | |
| VideoChat-TPOModel Scale=7B-8B, Training Paradigm=RL-based2025.12 | 25.2 | — | 34.5 | 19.3 | |
| VTimeLLMModel Scale=7B-8B, Training Paradigm=SFT-based2025.12 | 20.1 | 36.1 | 28.8 | 17.4 | |
| TimeChatModel Scale=7B-8B, Training Paradigm=SFT-based2025.12 | 11.6 | 40 | 14.4 | 7.6 |