Text-to-Video Generation Inference Speed on A100 80GB GPU (test)
11.2Latency (s)SimDA
Evaluation results
| Method | Links | |
|---|---|---|
| SimDAT2V Core Parameters (Billion)=0.88, Auto Encoder Parameters (Billion)=0.08, Text Encoder Parameters (Billion)=0.12, Overall Parameters (Billion)=1.08, Tuned Parameters (Billion)=0.0252023.08 | 11.2 | |
| LVDMT2V Core Parameters (Billion)=0.96, Auto Encoder Parameters (Billion)=0.08, Text Encoder Parameters (Billion)=0.12, Overall Parameters (Billion)=1.16, Tuned Parameters (Billion)=1.042023.08 | 21.23 | |
| Latent-ShiftT2V Core Parameters (Billion)=0.88, Auto Encoder Parameters (Billion)=0.08, Text Encoder Parameters (Billion)=0.58, Overall Parameters (Billion)=1.53, Tuned Parameters (Billion)=0.882023.08 | 23.4 | |
| Latent-VDMT2V Core Parameters (Billion)=0.92, Auto Encoder Parameters (Billion)=0.08, Text Encoder Parameters (Billion)=0.58, Overall Parameters (Billion)=1.58, Tuned Parameters (Billion)=0.922023.08 | 28.62 | |
| CogVideoT2V Core Parameters (Billion)=7.7, Auto Encoder Parameters (Billion)=0.10, Frame Interpolation Parameters (Billion)=7.7, Overall Parameters (Billion)=15.5, Tuned Parameters (Billion)=15.52023.08 | 434.53 |