Video Understanding on VideoMME, LongVideoBench, and MVBench Aggregate
60.3Average ScoreERA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ERABackbone=LLaVA-Video-7B, Token Retention Ratio=25%, Pruning Transfer Protocol=image-only pruning transferred to video inputs2026.06 | 60.3 | 97.5 | |
| LLaVA-OneVision-7BBackbone=LLaVA-OneVision-7B, Token Retention Ratio=100%2026.06 | 57.8 | 100 | |
| ERABackbone=LLaVA-OneVision-7B, Token Retention Ratio=25%, Pruning Transfer Protocol=image-only pruning transferred to video inputs2026.06 | 57.2 | 99 |