Video Generation on VBench Wan 5B 93x704x1280 2.2 (test)
65.04Aesthetic Quality (AQ)VMonarch
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| VMonarchModel=Wan2.2 5B, Resolution=93x704x1280, Kernel=VMonarch, Training Setting=Fine-tuned2026.01 | 65.04 | 96.68 | 59.72 | 66.69 | 94.2 | 92 | 205.5 | 92 | |
| Full AttentionModel=Wan2.2 5B, Resolution=93x704x1280, Kernel=FullAttn, Training Setting=Fine-tuned2026.01 | 64.99 | 96.67 | 62.5 | 65.44 | 94.56 | — | 352.4 | 123.7 | |
| VSAModel=Wan2.2 5B, Resolution=93x704x1280, Kernel=VSA, Training Setting=Fine-tuned2026.01 | 64.82 | 96.3 | 61.11 | 65.26 | 93.29 | 90 | 204.4 | 103 | |
| Full AttentionModel=Wan2.2 5B, Resolution=93x704x1280, Kernel=FullAttn, Training Setting=Training-Free2026.01 | 64.66 | 96.1 | 72.22 | 65.24 | 93.58 | — | 352.4 | 123.7 |