Streaming Video Understanding on RVS-Ego SVU (test)
59.4AccuracyProtoKV
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ProtoKVBackbone=Qwen2.5-VL-7B, Token budget |M|=24k2026.06 | 59.4 | 3.84 | |
| InfiniPot-VBackbone=Qwen2.5-VL-7B, Token budget |M|=24k2026.06 | 58.9 | 3.75 | |
| ProtoKVBackbone=LLaVA-OV-7B, Token budget |M|=24k2026.06 | 58.6 | 3.63 | |
| SWABackbone=Qwen2.5-VL-7B, Token budget |M|=24k2026.06 | 58.4 | 3.6 | |
| InfiniPot-VBackbone=LLaVA-OV-7B, Token budget |M|=24k2026.06 | 57.8 | 3.5 | |
| SWABackbone=LLaVA-OV-7B, Token budget |M|=24k2026.06 | 55.7 | 3.3 |