Update README.md
Browse files
README.md
CHANGED
|
@@ -6,7 +6,7 @@ Lightweight open-source reproduction of MagVitV2, fully aligned with the paper
|
|
| 6 |
* Training employs multi-resolution and dynamic-duration mixed training, allowing decoding of videos with arbitrary odd frames as long as GPU memory permits, demonstrating temporal extrapolation capability.
|
| 7 |
* The model is closely aligned with MagVitV2 but with reduced parameter, particularly in the lightweight Encoder, reducing the burden of caching VAE features.
|
| 8 |
|
| 9 |
-
[
|
| 10 |
|
| 11 |
## Contents
|
| 12 |
|
|
|
|
| 6 |
* Training employs multi-resolution and dynamic-duration mixed training, allowing decoding of videos with arbitrary odd frames as long as GPU memory permits, demonstrating temporal extrapolation capability.
|
| 7 |
* The model is closely aligned with MagVitV2 but with reduced parameter, particularly in the lightweight Encoder, reducing the burden of caching VAE features.
|
| 8 |
|
| 9 |
+
*[github_link: https://github.com/bornfly-detachment/asymmetric_magvitv2](https://github.com/bornfly-detachment/asymmetric_magvitv2)*
|
| 10 |
|
| 11 |
## Contents
|
| 12 |
|