Add pipeline tag
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
|
|
|
| 3 |
---
|
| 4 |
|
| 5 |
# Delving into Latent Spectral Biasing of Video VAEs for Superior Diffusability
|
|
@@ -12,7 +13,7 @@ Most existing video VAEs prioritize reconstruction fidelity, often overlooking t
|
|
| 12 |
downstream diffusion training. Our research identifies properties of video VAE latent spaces that facilitate diffusion
|
| 13 |
training through statistical analysis of VAE latents. Our key finding is that biased, rather than uniform, spectra lead
|
| 14 |
to improved diffusability. Motivated by this, we introduce **SSVAE (Spectral-Structured VAE)**, which optimizes the *
|
| 15 |
-
*spectral properties** of the latent space to enhance its **"Diffusability"**.
|
| 16 |
|
| 17 |
<div align="center">
|
| 18 |
<img src="https://raw.githubusercontent.com/zai-org/SSVAE/refs/heads/main/assets/figs/teaser.png" alt="Figure 1" width="400">
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
+
pipeline_tag: other
|
| 4 |
---
|
| 5 |
|
| 6 |
# Delving into Latent Spectral Biasing of Video VAEs for Superior Diffusability
|
|
|
|
| 13 |
downstream diffusion training. Our research identifies properties of video VAE latent spaces that facilitate diffusion
|
| 14 |
training through statistical analysis of VAE latents. Our key finding is that biased, rather than uniform, spectra lead
|
| 15 |
to improved diffusability. Motivated by this, we introduce **SSVAE (Spectral-Structured VAE)**, which optimizes the *
|
| 16 |
+
**spectral properties** of the latent space to enhance its **"Diffusability"**.
|
| 17 |
|
| 18 |
<div align="center">
|
| 19 |
<img src="https://raw.githubusercontent.com/zai-org/SSVAE/refs/heads/main/assets/figs/teaser.png" alt="Figure 1" width="400">
|