Long-context Video Understanding on VideoMME Sub (test)
74ScoreQwen3-VL-4B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-VL-4BModel=Qwen3-VL-4B, Pruning Strength=None2026.03 | 74 | |
| InternVL3.5-8BModel=InternVL3.5-8B, Pruning Strength=None2026.03 | 68.6 | |
| STTSModel=STTS, Pruning Strength=30%2026.03 | 68.5 | |
| BaselineModel=Baseline, Pruning Strength=0%2026.03 | 67.6 | |
| STTSModel=STTS, Pruning Strength=40%2026.03 | 67.4 | |
| STTSModel=STTS, Pruning Strength=50%2026.03 | 67.2 | |
| PLM-8BModel=PLM-8B, Pruning Strength=None2026.03 | 65.4 |