Unverified50% confidenceFactExact time
D3PM（Austin et al., 2021）将扩散过程推广到离散状态空间，通过转移矩阵描述 token 在离散状态间的随机跳转
1
Sources
50%
Confidence
Long-term
Relevance
8/22/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedThe Latent Diffusion Model was proposed by Rombach et al. in 2022, combining the diffusion process with the latent space of a VAE, reducing the computational dimensions from 512×512×3 to 64×64×4.66% similarUnverifiedDiffusionGemma 大概率采用掩码扩散方案，前向过程逐步将 token 替换为 [MASK] 标记60% similarUnverifiedThe basic workflow of spectral pooling consists of three steps: DFT transformation, spectral cropping to retain the central low-frequency components, and inverse Fourier transform for reconstruction.57% similarUnverifiedThe Latent Diffusion Model open-sourced by Stability AI moved the diffusion process into a low-dimensional latent space, reducing computational costs by an order of magnitude and giving rise to Stable Diffusion.55% similarVerifiedStable Diffusion compresses the diffusion process from pixel space to latent space, implemented by a Variational Autoencoder (VAE).55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/788421API
curl https://kongchang.com/api/v1/knowledge/claims/788421MCP
get_claim(id=788421)