Multi-modal Evaluation on Seed-Bench
0.79Seed ScoreVanilla
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VanillaModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.79 | 100 | |
| ToFuModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.788 | 100.75 | |
| ToMeModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.782 | 100.04 | |
| LightKVModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.78 | 101.37 | |
| PiToMeModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.774 | 100.24 | |
| FastVModel=Qwen2.5-VL-7B-Instruct, Vision token retention rate=55%2026.05 | 0.744 | 98.77 |