Video Extrapolation on 5 long descriptive prompts 2x / 165f train-free extrapolation Wan 2.1 1.3B
64.1VQevalLVSA (SDPA)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LVSA (SDPA)Backbone=Wan 2.1 1.3B, Attention Backend=SDPA2026.05 | 64.1 | 502 | — | |
| LVSA-FIBackbone=Wan 2.1 1.3B, Attention Backend=FlashInfer2026.05 | 63.7 | 395 | — | |
| UltraViCoBackbone=Wan 2.1 1.3B2026.05 | 62.1 | 741 | — | |
| RIFLExBackbone=Wan 2.1 1.3B2026.05 | 57.8 | 564 | — | |
| DenseBackbone=Wan 2.1 1.3B2026.05 | 57.2 | 566 | — | |
| LVSA-FI speedup vs. DenseBackbone=Wan 2.1 1.3B2026.05 | — | — | 1.43 | |
| LVSA-FI speedup vs. RIFLExBackbone=Wan 2.1 1.3B2026.05 | — | — | 1.43 | |
| LVSA-FI speedup vs. UltraViCoBackbone=Wan 2.1 1.3B2026.05 | — | — | 1.88 |