Video Extrapolation on 5 long descriptive prompts 4x / 333f train-free extrapolation Wan 2.1 1.3B
62.4VQEval ScoreLVSA (SDPA)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LVSA (SDPA)Backbone=Wan 2.1 1.3B, Attention Backend=SDPA2026.05 | 62.4 | 1.021 | — | |
| LVSA-FIBackbone=Wan 2.1 1.3B, Attention Backend=FlashInfer2026.05 | 62.3 | 0.802 | — | |
| UltraViCoBackbone=Wan 2.1 1.3B2026.05 | 58.8 | 2.621 | — | |
| RIFLExBackbone=Wan 2.1 1.3B2026.05 | 53.6 | 1.931 | — | |
| DenseBackbone=Wan 2.1 1.3B2026.05 | 52.4 | 1.93 | — | |
| LVSA-FI speedup vs. DenseBackbone=Wan 2.1 1.3B2026.05 | — | — | 2.4 | |
| LVSA-FI speedup vs. RIFLExBackbone=Wan 2.1 1.3B2026.05 | — | — | 2.41 | |
| LVSA-FI speedup vs. UltraViCoBackbone=Wan 2.1 1.3B2026.05 | — | — | 3.27 |