Unverified50% confidenceFactExact time
MDLM（Sahoo et al., 2024）专注于吸收态扩散，前向过程只将 token 替换为 [MASK]
1
Sources
50%
Confidence
Long-term
Relevance
8/22/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedDiffusionGemma 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66% similarUnverifiedD3PM（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53% similarUnverifiedGPTQ was proposed in 2022 and extends the OBS framework from pruning to quantization, using the inverse of the Hessian matrix to optimally adjust unquantized weights to compensate for quantization error, with efficient updates via Cholesky decomposition.52% similarVerifiedThe 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 64×64×4.52% similarUnverified长文档处理中，KV Cache机制导致显存占用随上下文长度近似平方级增长51% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/788422API
curl https://kongchang.com/api/v1/knowledge/claims/788422MCP
get_claim(id=788422)