Video Generation on HunyuanVideo 13B (test)
0.184CLIPSIMFull Prec.
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Full Prec.#Bits (W/A)=16/16, Density=100%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.184 | 81.23 | 0 | — | — | — | 1,264 | 1 | |
| QuantSparse#Bits (W/A)=4/8, Density=15%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.184 | 81.19 | 0.016 | 20.88 | 0.678 | 0.273 | 671 | 1.88 | |
| QuantSparse#Bits (W/A)=4/8, Density=25%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.183 | 79.05 | 0.014 | 20.86 | 0.675 | 0.272 | 731 | 1.73 | |
| Q-VDiT#Bits (W/A)=4/8, Density=100%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.182 | 67.95 | 1.168 | 16.85 | 0.605 | 0.461 | 1,155 | 1.09 | |
| Q-VDiT+SVG#Bits (W/A)=4/8, Density=25%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.182 | 70.99 | 1.379 | 16.71 | 0.595 | 0.458 | 743 | 1.7 | |
| Q-VDiT+SVG#Bits (W/A)=4/8, Density=15%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.182 | 76.3 | 1.393 | 16.66 | 0.591 | 0.46 | 687 | 1.84 | |
| ViDiT-Q#Bits (W/A)=4/8, Density=100%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.181 | 49.82 | 1.254 | 15.75 | 0.534 | 0.489 | 1,149 | 1.1 | |
| QuaRot+SVG#Bits (W/A)=4/8, Density=25%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.181 | 43.34 | 0.9 | 15.39 | 0.497 | 0.53 | 731 | 1.73 | |
| QuaRot+SVG#Bits (W/A)=4/8, Density=15%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.181 | 41.4 | 1.004 | 15.34 | 0.494 | 0.536 | 671 | 1.88 | |
| QuaRot#Bits (W/A)=4/8, Density=100%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.18 | 42.89 | 0.708 | 15.46 | 0.502 | 0.528 | 1,149 | 1.1 | |
| SmoothQuant#Bits (W/A)=4/8, Density=100%, CFG=6.0, Resolution=720 × 1280p, frames=602025.09 | 0.178 | 42.21 | 1.194 | 15.44 | 0.479 | 0.583 | 1,149 | 1.1 |