Unverified50% confidenceFactExact time
Discrete flow matching is technically an extension of diffusion models into the discrete domain, migrating the denoising approach from the continuous pixel space to the discrete token space of text.
1
Sources
50%
Confidence
Long-term
Relevance
9/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedStable Diffusion 基于去噪扩散概率模型（DDPM）理论，将扩散过程从像素空间转移到低维潜在空间，由VAE负责编解码，CLIP模型将文本提示编码为语义向量76% similarUnverified文本扩散模型主要通过两条路径解决离散token问题：在连续嵌入空间执行扩散，或定义离散马尔可夫转移矩阵直接在token空间操作75% similarUnverified离散扩散语言模型主流方案包括：将token映射到连续嵌入空间进行扩散（如Diffusion-LM），或设计离散扩散过程用掩码模拟加噪（如D3PM、MDLM）75% similarUnverifiedDiffusion-LM uses learnable embeddings rather than fixed embeddings because the geometric structure of the embedding space needs to mutually adapt to the denoising dynamics of the diffusion process.73% similarUnverified当假词编码向量落在流形之外时，扩散模型去噪过程会将结果拉回到最近的流形表面，因此即使输入无意义输出也看起来连贯73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/927900API
curl https://kongchang.com/api/v1/knowledge/claims/927900MCP
get_claim(id=927900)