Video Generation on VBench and Vision Reward Mixkit 2000 videos
97.15SCTop-K
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Top-KEvaluation Protocol=Tuning-based, Strategy=SALAD2026.01 | 97.15 | 96.27 | 64.76 | 25.83 | 0.107 | 90 | 1.04 | |
| Top-KEvaluation Protocol=Tuning-based, Strategy=LoRA2026.01 | 96.84 | 96.1 | 66.03 | 24.02 | 0.058 | 80 | 0.95 | |
| Spatial-TemporalEvaluation Protocol=Tuning-based, Strategy=SALAD2026.01 | 96.54 | 96.37 | 66.09 | 25.55 | 0.092 | 90 | 1.72 | |
| Original (Full Attention)Evaluation Protocol=Baseline2026.01 | 95.88 | 96.17 | 65.93 | 25.31 | 0.109 | 0 | — | |
| Spatial-TemporalEvaluation Protocol=Tuning-based, Strategy=LoRA2026.01 | 94.96 | 95.29 | 65.27 | 25.39 | 0.072 | 63 | 1.52 | |
| PAROAttenEvaluation Protocol=Training-free2026.01 | 93.67 | 94.87 | 64.88 | 25.01 | 0.064 | 56 | 1.2 | |
| SVG2Evaluation Protocol=Training-free2026.01 | 90.75 | 94.31 | 63.07 | 25.08 | 0.031 | 45 | 1.48 | |
| SLAEvaluation Protocol=Tuning-based2026.01 | 83.77 | 96.26 | 25.82 | 3.6 | -0.205 | 90 | 1.42 | |
| Spatial-TemporalEvaluation Protocol=Training-free2026.01 | 80.14 | 93.96 | 42.07 | 7.98 | -0.2 | 63 | 1.52 |